DIGITAL IMAGING for PHOTOGRAPHIC COLLECTIONS Foundations for Technical Standards seonarasin Franziska 8. Frey James M. ReillyThe bigest litrature Website of the world WWW.QALAMURDU.COM we rd COMfae CONTENTS ABOUT 'THE SECOND EDITION. u.-seesseeeeees eeseeeeeeetinneee a ACKNOWLEDGMENTS wesc a eae se crane? FOREWORD. — mse eee mse meen Ut INTRODUCTION... ‘Tue Prases oF THe Prorecr BACKGROUND OF THE PROJECT......:scssee0 RLG Tecnwcat bwces Test Prosacr 5 Evaluation of Image Quality 5 Results from Image Comparisons in the RLO Technical Images Test Project 5 Wrere We Starrep 6 LrvereroRe RESEARCH AND PREPARATION OF SAMPLES 7 Significant Projects for This Study 7 ‘The Salient Points 8 BUILDING THE IMAGE QUALITY FRAMEWORK... ‘Wear Is nace Quatsny? 10 icra, Ine QuaLery aND Frs IMeLicxrions FoR av Inaceye Praascr 11 Bonpive Visuat. Lrreracy 13 Sosecrve Iace Quatiry Evauerion 13 Dernirron oF THE Parwerers ror Exatuarvc Teceicat, Picromtat. Quarry 14 ‘The Role of Targets in Evaluation of the Image Quality Parameters 15 ‘The Parameters oF THE IMsce QUALETY FRAMEWORK qv ‘Tone Reproduction 17 ‘Tangots to Use 19 Detail and Edge Reproduction (Resolution) 20 ‘What is Digital Resolution? pa How Is Digital Resolution Measured? 2 ‘The Modulation Transfer Function (MTF) 2 ‘Targets to Use 4 Noise 26 ‘Target to Use EaColor Reproduction oT Pictorial Rendering Intent 28 Choosing a Color Space 30 New Tools and Developments 30 Image Artifacts 32 Sermve Ur lnneme Srstems 32 Monitor Calibration 32 Drerou. Masrer ap Dernsrnves 33 Quanrry AND Process Controt. 33 Benchmarking, Scanner Systems 34 Reproduction Qualities and Characteristics of the Digital Master and Derivatives 34 Functional Qualities and Characteristics of the Digital Master and Derivatives 35 Documentation of the Imaging Process—Administrative Metadata 35 Ihwce Processine 35 Processing for Archiving 36 Processing for Access 36 Dato: Compression 36 ageless andl Visnally Lossless Compression 36 Lossy Compression 37 ‘THE CONFERENCE...... CONCLUSIONS REFERENCES. BIBLIOGRAPHY - SELECTED INTERNET RESOURCES...Fie ABOUT THE SECOND EDITION Close to seven years have paseed since the first printing of thie booklet. While many technological advances have ‘been made in imaging in the meantime, the fundamentals described in this booklot still hold true. The field is closer than ever to getting all the necessary standards and tools in place, hut there are etill many challenges to overcome to build the seamless workflows needed. We have updated all the cited internet resources and references to reflect the changes that have happened in the years since the firet edition. We have, however, left the main text unchanged and are convinced that it remains a valuable resource. Franziska Trey and James Reilly Rochester, NY July 2006 MENTS me ACKNOWLED¢ ‘The Image Permanence Institute gratefully acknowledges the support of the Division of Preservation and Access of the National Endowment for the Humanities for this project, We would lke to thank the American Memory “Team of the Library of Congress, specifically Carl Fleis- chhauer and Phil Michel, who made it possible to translate our ideas into the actual project. Special thanks go to Sabine Siisstrunk, EPFL, Lausanne, Switzerland; Rudalf Gechwind, University of Basel, Baeel, Switzerland; Asher Gelbart; Jack Holm, HP Labs, Palo Alto; Stave Puglia, National Archives; and Stephen Chapman, Harvard University Library. Thanks to Karen Santoro for design, editing, and illustrations, This work would not have been possible without the many individuals in the field working on digital projects who willingly shared their experiencesFie FOREWORD ‘There is a well-known truth among photographers that, copy photography—taking a picture of a pieture—is tech- nically more difficult than making a portrait or a land- scape. It is all too easy in copy work to alter the contrast or Jose the fine details of the original. As cultural institutions hegin to make digital copies of their photograph callec- tions, they are learning a variant of the old truth: itis as difficult to make good digital eopies as good photographic cones. The goal of the NEH-sponsared project that made this publication possible was to offer come guidance to libranes, archives, and museums in their efforts to convert photographic collections to digital form. Specifically, we sought to identify the key issues affecting image cpual- ity, clarify the choices facing every digitizing project, and explore ways to measure digital image quality. Along the way, we learned that one of the most important and difficult questions to answer is what level of quality is really needed in digital image collections Tn the hands of expert operators, today’s beet digital imag- ing harclware is capable of capturing aff the information in photographic originals. Such a high quality standard produces the most versatile digital images but requires the storage and manipulation of huge file. Alower quality standard produces more manageable files but often limits the utility of the files for such demanding uses ae publica tion or exhibition. Selecting the appropriate quality level will always depend on careful analysis of the desired uses of the images in the near and long term. In the project and in this publication, we have sought to clarify not only the qualitative aspects of quality choices, Inut aleo the technical and quantitative, Measurements that ensure adequate capture of detail and contract are in some ways easier anc more accurate in digital imag- ing than in conventional photography because they can be done in software. Only when off-the-shelf software for thie purpoee becomes available can the full promize of digital imaging for institutional photograph collections be realized. Franziska Prey and James Reilly Rochester, NY September 1999Fie INTRODUCTION ‘The grant project, Digital Imaging, for Photographic Collections: Roundations for Technical Standards wes & two-year research study that investigated the use af digital imaging in linaries and archives. “There are no guidelines ar accepted standarcs for determining the level of image quality required in the creation of digital image databases for photographie collections, Although clear imaging standards are not yet fully in-place, and it is difficult to plan in an atmosphere of technological uncertainty, there are come basic rules that can he followed to minimize unexpected recults.? Within this project we began to develop an “image quality framework” that will help with planning parts of digi- tal-imaging projects, Our framework is not meant to be complete and ready to use; more work on various levels is needed. However, fruitful discussione were begun through this work, and we can see some of ouricleas heing taken forward in other initiatives ‘The materials that make up photographs (ilver or dyes as image-forming materials; paper, celluloid, or other plastics az hase materials; and gelatin, albumen, or collo~ dion az binders) are not chemically stable. Environmental influences such as light, chemical agents, heat, humility, and storage conditions affect and destroy photographic materials, In a general way, the life span of photosraphe can be extencled only by appropriate etorage at low temperatures and low humidity: Storage under controlled climatic canditions does not prevent decay, however; it results, at hest, in a significant slowdown, Photographic collectione are facing a dilemma: on one hand, photo- graphic documents rust be stored under correct elinnatie conditione, and, on the other hand, itis often necessary to have cquick access ta them. Frecquently,in this situation, an umzatisfactory compromsize is mace.* ‘The long-term preservation of photographie images is always a very dermanding task. The principles of eecure prezervation for digital data are fundamentally different from those for traditional analogue data** First, in tracl- tional preservation there is a gradual decay of image «ual- ity, while digital image data either can he real accurately‘THE ADVANTAGES OF DIGITAL INFORMATION + Digital data represents asymbolic description of the originals; it can ‘be compared to the invention of writing, + Digital information ean be copied without loss. + Active sharing of digital images is easily: possible + New viewing experiences are possible by browsing through a collection without pre-selection by another individual. This will allow a completely different type of intellectual access to pictorial information. Digital Imaging for Photogray lections: F or, in most eases, cannot be read at all. Secondly, every analogue duplication process results in a deterioration of the cruality of the copy, while the duplioation of the digital image data is possible with no loss at al In an idealized traditional archive, the images should be stored under optirnal climatic eonditione and never ‘touched again. As a consecquance, access ta the images is severely hindered while clecay is only slowed down. A digital archive has to follow a diffrent eteategy. The safe keeping of digital data requires an active and reaular maintenance of the data. The data have to be copied to now media hefore they become unreadable. Since infor- mation technology is evalving rapidly, the lifetime of hoth software and hardware formats is generally shorter than +the lifetime of the reocrding media. However, since the digital data can be copied without loes, even if the media type and harelware change, the image isin a “frozen” state, and the decay has completely stopped. “The main difference between a traditional archive and a digital archive is that the traditional archiving ap- proach is a passive one with images being touched as little as possible, Often, however, this works only in theary. Ifa document is known to be available, itis likely to be used. ‘Therefore, in practice, we see an increased hanclling of original documents as soon ae they become available in digital form. The future will chow whether a good enough digitized copy will reduce this behavior ‘The digital archive needs an active approach in which the digital data (and the media they are recardled on) is monitored continually. This constant monitoring and copying can be achieved with a very high degree of amttomation ‘One of the big isenes that institutions should consider ‘prior to implementing a project is the anticipated use of their digital image collections, Will the images he made accessible on a stand-alone workstation or via the World Wide Web? Will they be used for printing reproductions? ‘What size will the prints be? Are there restrictions on ac- cess that must be honored? These are only a few of the questions that have to be answered befare a digitization ‘project starts, ‘There is a growing cansensue within the preservation community that a number of image files must he created for every photograph to meet a range of uses, First, an nations for Technical StandardDIGITAL MASTER VERSUS DERIVATIVES The digital master is the file that isarchivecl It represents the high- est quality file that has been digi- tized. Since this ix the information that is supposed to survive and be taken into the future, the main is- sues in creating the digital master relate to longevity and quality. The derivatives are the files for daily use. Speed of access andl transmission and suitability for certain purposes are the main ts- sues to consider in the creation of derivative files “archive” or master image should be created. It should. contain a brightness resolution greater than eight bits per channel, it should not be treated for any specific output in mind, and it chould be uncompressed or compreseed in a loesloss manner. From this archive file various access files can be produced as needed to meet apecific uses. "The following three examples illustrate the ways in which the intended use drives decisions regarding digital image quality: * The digital image is used only as a visual reference in an electronic database. The required digital innage quality is low, in torms of both epatial and brightness resolution content. ‘The display is usually limited to a screen or a low- reealution print device. Thumbnail image size for screen viewing usually does not exceed a width of approximately 250 pixele, If an additional, larger image eize is desired for low-resolution proview- ing on a print device or larger viewing on screen, pixel dimensions of 1024 x 768 are sufficient for most applications. Exact color reproduction is not critical, Additionally, images can be compressed to save storage space and delivery time. ‘The digital image is used for reproduction. ‘The requirements for the digitizing eyetem will dopend on the definition of the desired reproduc- tion, Limiting output to certain epatial dimen- sions will facilitate the decision-making process For example, if the output is limited to an 8 x 10 hard copy at a reeolution of 300 dots per inch (dpi), the dimensions of the digital file need not exceed 2.400 x 3,000 pixels. Similarly, decisions regarding tonal reproduction are facilitated when modestly sized reproduotions in print are the goal of digitization, Currently, most digitizing systems will allow only an eight-hit-per-color output. This is in moet cases, a perceptual, not a colorimetric, rendering of the original. It is important to note that if theze colors are not mapped correctly, the dligital file may not always replicate the tone and color of the original ‘The digital image represents a “replace ment” of the original in termes of spatial and4 Digital Imaging, + Photog tonal information content. This goal ie the most challenging to achieve given today’s digitizing technologice and the cost involved. The informa- tion content in terme of pixel equivalency varies from original to original. Itis defined not only by film format but also by emulsion type, shooting conditions, and processing techniques, Additional ly, eight-bit-per-color digital capture might be ad- equate for visual representation on today’s output devices, but it might not he sufficient to represent all the tonal subtleties of the original. Ultimately, “information content” has to be defined, whether based on human perception, the physical proper- ties of the original, or a combination of both, ‘Tue Prasks oF THE ProyecT ‘The first phase of the project involved searching the most recent technical literature, making conneetions with other peaple and projecte, defining and planning the techni- cal image choices to he explored, setting up an imaging workstation, and having a closer look at the eample images, that hadl been created in another initiative, The rapid and ongoing changes in the field made this a constant task throughout the project ‘The main outcome of the second phase was a frame- work to define eubjective and ohjectiva image parameters. Because thoze working with the new imaging technologies are only now beginning to understand all the associated issues, definitions of the parameters and tools to measure them are not readily available, IPI has defined some of the parameters, focusing on the materials found in photo- graphic collections ‘The colloquium entitled Digitizing Photographie Collections —Where Are We Now? What Does The Future Hold? took place June 7-9, 1997, at Rochester Institute of Technology (RIT), The event received a lot of atte tion and brought to Rochester over 120 attendees and 20 speakers from around the world nations for Technical StandardFigure 1. The test images of the RLG technical image project filled « number of binders, since the choices for duplication are rumerous Figure 2, Image quality coaluation twas performed by subjectively comparina all ofthe different prints to each other fa BACKGROUND OF THE PROJECT RLG Technica Imaces Test Prorecr ‘This project was based on the Technical hmages eet Proj- cot started by a task force of the Research Libraries Group, in 1992.? The project was designed to explore how image qquality is affected by various choices in image capture, digplay, compression, and output. There are many ways to create and view digital images. The task force felt that, although there was no “best” way for every collection or every project, it would be helpful to define and explore a finite range of practical choices. ‘The RLG Technical Images Test Project was able to achieve some success in clarifying image quality issues for the initial step of image capture. It also became clear, however, that the given taek was much more complex and consumed more time and resources than anticipated. Evaluation of Image Quality Fourteen images representing a range of photographie procesces and sizes of originale were picked from the IPI study collection, From theze photographe were prociced negatives, positives, color internegatives, or duplicate transparencies (Figure 1). The same photographs were then used to generate positive photographie prints, in either color or black and white, The haete method of comparing «quality reeulting from the various intermediate formats and film typee wae to lay all the prints out on a table under controlled lighting conditions (Figure 2). With the naked, eye, with a loupe, or with a stereo microscope, prints were compared to each other two at a time. The sharpness of the same celected detaile in each print were compared, as were graininess and smoothness of tone, For the color originals and the nineteenth-century processes, color fidelity was also ‘important, hut it was recognized that there is an extra layer of difficulty in controlling and judging color fidelity Results from Image Comparisons in the RLG Technical Images Test Project ‘The image comparison exercise chowed us that the first important decision to he made at the digital capture stageB Figure 3. Scanning from original or intermediate? Top image is « contact print fiom an original 8.x 1 negative. The arrow points to a detatl chosen for comparison. Image A shows the selected detail ‘from a scan of the original; image B shows the ‘sume detail from a sean of a 85mm intermediate Digital Imaging for Photogray lections: F is whether digitization chould be done directly from the original photographs themselves or from photographie copies, alsa knawn az photographic intermediates (Fig ure 3), There are advantages and disadvantages to hoth approaches. Because every generation of photographic copying involves come quality loes, using intermediates immediately implies some decrease in cquality. Intermedi- ates may aleo have other uses, however: for example, they raight serve a2 masters for photographic reference copies or as preservation exrrogatee ‘This leads to the question of whether the negative or the print ehould be used for digitization, assuming both are available. As stated ahove, cquality will always he best if the firet generation of an image, ic., the negative, is used. However, it can happen, mainly in the demain of fine-art photography, that there are hig differences between the negative and the print. The artist often spends a lot of time in the darkroom creating his prints, The reeulte of all this work are loct if the negative, rather than the print, is scanned. The outeome of the digitization will be disap- pointing, "Therefore, for fine art photographs, seanning from the print ie often recommended. Each caze must be looked at ceparately, however: ‘One of the mest important lossons leamed in this exercise was that many of the original photographs in institutional collections have truly outstanding image cpual- ity. Even using 4.x 5 duplication film, excellent lenses, and tho skill and experience of many years of photagraphic duplication, there was still a quite noticeable lass of image quality in the copying of an original & x 10 negative. Other images showed the same results, to varying degrees. This high level of image quality in original photographs sets @ very high standard for successful reformatting projects, whether conventional or digital. We must be careful not to “reformat” the quality out of aur image collections in the name of faster access. Instead, we must learn what it takes to bring that cuuality forward for future generations or at least to know that we are failing to do eo. Waere We StarTep ‘The RLG Technical Images Test Project showed us that the question of image quality cannot he answered in a linear fashion, Scanning, processing, and outputting im-Figure 5. wovw-corbis.com ages involve many different parameters that affect image quality. In any large digital imaging effort, objective pa- rameters and ways to control them must he defined at the outset, [PI's goal was to attempt to quantify image cpuslity by building an image quality framework, We learned a great deal from this procese—rnost importantly, that there is a Int more work to be done. Calibration issues need to he salved before image quality can he evaluated on a monitor. Methods for testing the capabilities of, or benchmarking, eysterns must be devised, Most critical of all, a common language in which all in- volved parties can corareunicate needs to be developed, LITERATURE RESEARCH AND PREPARATION OF SAMPLES ‘The first phase of the project consieted of examining digital imaging projects in the libraries and archives field. This ‘was accomplished thraugh personal contact with people at institutions conducting such projects, watching various dliscuceion lists and the growing number of museum sites on the Internet, and rezearching newaletters and scientific journal: in the field. Because everything ie changing eo quickly in the digital field, literature research and exami- nation of digital projects ware ongoing tasks, proceeding hand-in-hand with monitoring and leaming about new technologies." Significant Projects for This Study + National Digital Library Project (NDLP) of the Library of Congress. The Library plans to convert as many as five million of its more than one hundred million items into digital form before the year 2000. The materials to be converted inchide books and pamphlets, manuscripts, prints and photographs, motion pictures, and sound recordings: NDLP used part of our image quality approach in their request for proposals and in the process control for their current scanning projects (Figme 4) Corbis Corporation. Corbis is one of the big gest digital stock agencies in the world. With the incorporation of the Bettmann Archives, Cothis faces new problems similar to the ones archives and libraries have, due to the fact that there ie aorn Figure 7.IT10 task group forthe characteiza- tion of scanners. The frst work item deals tcith resolution measurements for seanners. Figure 8. ww amico.org (In June of 2006, the members of the Art Muscum Image Consortium voted to dissolve their collaboration.) 8 Digital Imaging for Photogra lections: F whale new range of materials that will have to be scanned. Contact with Carhie gave-us an interest ing insight into some of the issues to be faced hy a digital etock agency, a company whose ultienate goal is to make rnoney with digitized photographe (Figure 5) * Electronic Still Photography Standards Group (ANSI IT10), The scope of this commit- toe inchucles epecifying storage media, device inter- faces, and image formats for electronio still picture imaging, The scope alsa inchides standardizing, measurement methods, performanes ratings for devices and media, and definitions of technical terms, Within the committee, a new tack group for characterization of ecanners has been formed. Some of the findings of the evrrent project are being used as a basis for a standard in this area. ‘The cumulative experience that was brought into this project is very valuable and will speed up the yublication of much-needed dacuments (Figures 6 and 7) + AMICO (Ari Museum Image Consortium). RIT, together with the Image Permanence Insti- tute, has been selected to he among the partici- pants of the University Testhed Project. AMICO has been formed by twenty-three of the largest art museums in North America. The mission of this nonprofit organization is to make a library of digital documentation of art available under educational license (Figure 8). The Salient Points IPPs ongoing review of work in the field showed a grow- ing awarenees of the consplexity of undertaking a digital conversion project. In addition to the creation of a digital image database, which by iteelf brings up numerous problems, maintenance of the database over time must be considered. Questions regarding the permanence af the storage media, acceesibilty, quality control, and constant updating are only a few of thoze that rust be addressed We also saw proof that many of the problems arising from the need to scan for an unknown future use are not yet solved andl that there is a great deal of uncertainty‘THE NEED FOR COMMUNICATION Tits up to libraries and archives totell the hardware and software industries evactly what they need, but before a fruitful dialogue can take place a common language must be developed. Some firddings from thia project concerning spatial resolution have been used as a basis for the ISO Standard 16067-12003, Electronic ecannere for photo- graphic images —Spatial resalu- tion measurements: Part 1, Scan- ners for reflective media about how to proceed. Thase responsible for some af the big digital reformatting projects report the came problem: rapid changes in the technology make it difficult ta choose the best time to set up a reformatting poliey that will not be outdated tornorrow. ‘The lack of communication between the technical field and institutions remains a formidable obstacle, It cammot be emphasized enough that ifinstitutione fail to comamumicate their needs to the hardware and software industries, they will not get the toole they need for their epecial applications. Archives and libraries should know that they are involved in creating the new standards, It can be seen toclay that whoever is first on the market with anew product is creating a de facto digital technology standard for competitors. Furthermore, time ta create new standard is very chort; industry will not wait years to introduce a product simply because people cannot agree ona certain issue, Both institutions and industry are inter~ ested in a dialogue, but there is no common language.”*" ‘The exponential growth and use of the Internet has: raised a. whole new range of questions and problems that will have to he solved, but the Internet is aleo a great information resource Adigital project cannot be looked at az a linear pro- cees, in which, one task follows the other. Rather, it has to be looked at as a complex structure of interrelated tasks in which each decision has an influence on another one. ‘Tho first etep in penetrating this complex structure is to thoroughly understand each eingle step and find metrics to qualify it. Once this is done, the separate entities can he put together in contest, We are still in the first round of this process, but with the benefit of all the experience gathered, from the various digital projects in the field, we are reach- ing the point at which the complex system can he looked at as a whole Background of the Project 9r Ph ssraphic C lection fe BUILDING THE IMAGE QUALITY FRAMEWORK Wuar Is Imace Quaurry? According to The Focal Eneyelopedia of Photography, [ffhe basic purpose of a photograph is to reprocluce an image. One of the three hasic attributes ofa reproduction image is the reproduction ofthe tones of the image. Also of importance are the definition af the image (¢he reproduction of eeges and detail and the saruonnt of noize in the inaage) and the color reproduction. It is convenient to deal with these attributes when evaluating an image.” Image quality is usually separated into two classes * Objective image quality is evaluated through physical measurements of image properties. Historically, the definition of image quality has emphasized image physics (physical image pa~ rameters), or objective image evaluation. + Subjective image quality is evaluated through judgment by human observers. Stimuli that do not have any meaeurable physical quantities can he evaluated by using peychometrie sealing test methods. The stimuli are rated according to the reaction they produce on human obeerv- ere, Peychometric methods give indications about reponse differences, Psychophysical scaling tools to measure eubjective image quality have he available only for the last 25 to 35 years (Quantification of image quality for the new imaging, technologies is a recent development.” The theoretical knowledge and understanding of the different param- ters that are invelved is available now. Still missing for the practitioner of digital inaaging are targets and tools to objectively measure image culty. These tools are available only within the companies that mamufacture imaging systems and are wsedl mostly to benohark the companies’ own systeme, Furthermore, in most cases, the systems heing used for digital imaging projects are open systems, which means that they include modules from dif-ferent manufacturers, Therefore, the averall image cuuality performance of a system cannot he predicted from the manufacturers” specifications, sinoe the different compa- nnts influence each other It should be keptin mind that scanning for an archive is different from ecanning for preprees purposes. In the latter case, the variables af the soanning process are well known, and the soanning parameters can he chosen ao- cordingly. If an image is scanned for archival purposes, the future use of the image is not known, and neither are technological changes that will have taken place a fow years frora now. his leads to the conclusion that decisione concerning the quality af archival image scans are very cxitical As soon at various conferences, this isa relatively new concept for both the inetitutions and the technical field, and it will take come work to help both sides understand whore the probleme are, This ie « topic for more research. ‘The ANSI Standards task group for the characterization of scanners has contributed to the technical community's awareness of the issue Image quality evaluations are important at two dlif- ferent etages of a project: at the beginning, to benchmark the cystem that will be used, and later, to check the images that have been produced. **"7 Diertan IMace, QuALITY AND [1s IMPLICATIONS FOR AN TMaciNG PRojEcT ‘Thre are no guidelines or accepted standards for deter- raining the level of image quality required in the creation of digital irnage databases for access or preservation of photographic collections, A a result, many institutions have already been disajspointed because their efforts hhave not lead to the results they were hoping for: Either the paramoters chozen for the digitization process were not thought through, or the technology changed after the project started. In the first case, the failures might have been prevented. No one knows what technology will be available in afew years, however, and the tack of choosing the right scanning parameters till needs to be researched, One problem is that we are currently at the beginning of the eyele of understanding image cquality for the new imaging technologies. Building the Image Quality FrameworOften, imaging projects start with ecanning text docu- ments, When thie type of material is scanned, quality is calculated according to a “Quality Index,” which is hased on published guidelines for microfilming (Figure 9) Quality Index is means for relating reaolue tion and text legibility, Whether it is used for microfilming or digital imazing, Qlis based on relating toxt legibility to system reeolution, ie, the ability to capture fine detail ‘There is no common tmit like character size that can ensure the semantic fidelity of images. The quantitative colutions that have evalved frem ecanning text therefore do nat have a logical or practical fit with capturing images” ‘This means that other waye have te be found for the evaluation of the quality of an image and for determining how much of the image information is conveyed in the digital reproduction. ‘The more one looks at image quality andl ways to clearly define it, the more parameters have to he taken inta account. We have tried to develop a balanced ap- proach that is not only usable in the archives and hbraries Figure 9. The main quality issue for reformatted world but aleo as complete as possible from an engineering, paper documents is legibility: The higher the ppi, point of view. the more details of the character can be resolved. In addition, when looking at image quality, the whole The needed ppi can be ealeulated for a known image processing chain has to be exarnined. Besides iesuees character height cred concerning the scanning system, [PI has looked at com- pression, file formats, image processing for various uses, and system calibration. One of the hig issues is that institutions will have to decide beforehand on the use of their digital images. This still creates a lot of questions and problems. Sometimes, how images will be used is not clear when a project starts, More often, however, institutions don’t take enough time to think about the potential use of the digital images Fim thermore, institutions often have unrealistic expectations about digital projects. Even if the goals have been carefully defined, costs may not he worked out accordingly, or goals may not match the available funds Although the ease of use of many digitizing «ystems has fostered the perception that scanning is “simple,” snccessfully digitizing a photographie collection requires as much experience as conventional reformatting.” Further, most of the available scanning technology is still based 12 Digital Imaging for Photographic Collections: Foundations for Technical Standard:HOW BIG IS THE COLLECTION? When choosing a digitizing sys- tem, bear in mind that approaches that work for a small number of images may not work for a large number of images. Fery: sophis- ticated systems can be set up ir a laboratory environment for a amall number of images. Large numbers of images need a well thought out workflow. Figure 10, How not to do it: comparison of the original printed reproduction and « digital image on the monitor. Tobe done correctly, subjective mage quolity evaluation requires a standardized environment with calibrated monitors and dim room illumination. on the model of immediate output on an existing output device with the original on hand during the reproduction process. Spatial resolution and color mapping are deter- mined hy the intended output devies. Depending on the quality criteria of the project, a more sophisticated system and more expertize by the operator are needed to success- fally digitize a collection in an archival environment where the future output device is not yet known. The charac- teristics of scanning devices euch as optical resolution, dynamic range, registration of color channele, hit-depth, noise characteristics, ancl quantization controle need to be carefully evaluated with consideration of the final use of the digital images When choosing the digitizing system, it alzo must be remembered that approaches that work for a small num- ber of images may not be suitable for the large number of images uzually found in collections, Burpine Visa Literacy Looking at images and judging their quality hae always ‘been a complex task. The viewer has to know what he/she is looking for The visual literacy required when looking at conventional images needs to be translated for digital images. Much more research is needed to enable us to fully understand the ways in which working with images on a monitor differ from working with original photographs. Supmecrve Ivace Quanrry Evacuation Inmost cases, the firet evaluation of a scanned image will bbe made by viewing it on a monitor (Figure 10). The viewer will decide whether the ima on the monitor fulfills the goals that have been stated at the beginning of the seanning project. This is important, because human, juelgment decides the final acceptability of an image. It should be emphasized that subjective quality contral must be executed on calibrated equipment, in an appropriate, standardized viewing environment. If tone and color are evaluated, it may be necessary to transform data to suit, display viewing conditions, ‘While the image ie viewed on the monitor, defeats such as dirt, “half images,” skew, and so on, can he detected. In addition, a target can ho used to check the registration of the three color channels for color seams. It Building the Image Quality Frameworr Ph ssraphic C lection is aleo important to check visual sharpness at this point Mechanical malfimotion of the scanner or limited depth of field could cause images to lack sharpness. DEFINITION OF THE PARAMETERS FOR EVALUATING Tecencan Picrorw Quarry ‘There are four main parameters ta consider when assess- ing the technical pietorial quality of an image. Due to the lack of off-the-shelf coftware and targets to evaluate the quality parameters, IPI had to create its own software tools and targets, As a result, two parameters, tone repro- duction and detail and edge production, became the focus of the project. * Tone reproduction. This refers to the degree to which an image conveys the luminance ranges of an original scene (or of an image to be repro- duced in case of reformatting). It is the single most important aspect of image quality Tone reproduc- tion ie the matching, modifying, or enhancing of ‘output tones relative to the tonee of the original document. Because all of the varied components of an imaging eyetem contribute to tone reprochuc- tion, itis often difficult to control * Detail and edge reproduction. Detail is defined as relatively emall-scale parts of a subject or the images of thoee parts in a photograph or other reproduction, In a portrait, detail may refer to in- dividual haire or pores in the skin, Edge reproduc tion refers to the ability of a process to reproduce sharp edges (the visual sharpness of an image’. Noise. Noize refers to random variations aseoci- ated! with detection and reproduction systems. In photography, the term granularity is used to cecribe the objective measure of density nonuni- formity that corresponds to the subjective concept of graininees. In electronic imaging, noige is the presence of unwanted energy fluctuation in the signal. This energy ie not related to the image signal and degrades it. Color reproduction. Several color reproduction intents can apply to a digital image. Perceptual intent, relative colorimetric intent, and absolutecolorimetric intent are the terms often associated with the International Color Gansortium (ICC), Perceptual intent is to create a pleasing image on a given medium uncer given viewing conditions, Relative colorimetric intent is to match, as closely as possible, the colors of the reproduction to the colors of the original, taking into account output media and viewing conditions, Absolute colorimet- ric intent ie to reproduce colore as exactly as pos- sible, independent of output media and viewing, conditions, ‘These parameters will be looked at in greater detail in a later section, The Role of Targets in Evaluation of the Image Quality. Parameters ‘Targets are a vital part of the image quality framework. ‘Ta be able to make objective measurements of each of the four parameters, different targets for different forms of images (e.g. prints, transparencies, ete.) are needed. To get reliable results, the targets should consiet of the same materials as those of the item that will be seanned—pho- tographic paper and film, After targets are scanned they are evaluated with a software program, Same software components exist as plug-ins to full-featured image brows- ers, others as stand-alone programe ‘Tangets can be incorporated into the work flow in various ways, Full versions of the targets might be scanned every few hundred images and then linked ta specific hatches af production files, or smaller versions of the targets might he included with every image. The chosen method will depend on the individual digital imaging project. As more institutions initiate digitization projects, hav- ing an objective tool to compare different scanning devices will he mare and mare important. Until now, seanner manufacturers usually have used their own saftware when evaluating and testing system For the current project, IPT examined approaches tak- en by other research projects in similar areas and looked carefully at a variety of targets, Some targets were alreacly available for other purposes and could he purchased; some had ta be custom-made, A cot of tangets with software to read them are available on the market (see page 45) Building the Image Quality Frameworr Ph ssraphic C lection ‘The use of objective measurements resulting from the target evaluation will be twofold. Some of the results, together with arlditional information like spectral eensitivi- tiee and details about the actual image proceseing chain, will he used to characterize the eoanning eystem. This assumes that spectral sensitivities are known and that a complete description of the image processing chain is at hand. These requirements are not often fulfilled, however, since scanner manufacturers are reluctant to provide this information. Other reeults of the target evaluation will be associated with each image file; this information will be used to perform data corrections later on as the images are processed for output or viewing, Therefore, standardized approaches and data forms are recquired for interchange- ability of the data In “Specifics of Imaging Practice,"M. Ester wrote: IF Tse shortcomings in what we are doing in documenting, images, they are traceable to the lack of standards in this area. We have responded to a practical need in our work, and have settled on the information we believe isimportant to record about production and the resulting image resource. These recording procedures have become stable overtime, but the data would become even more valuable if there was broad coramuity consensns on a preferred framework. Compatibility of image data from multiple sources and the potential to develop software around access to & common framework would be sorae of the advantages." New file formats like TIFF/EP* that inchice a large number of defined header tags will facilitate the stancard- ined storage of image attrilmte information (acministra- tive metadata}, which, in turn, will facilitate future image processing Applications do not yet support TIFE/EP, but it is important that collection managers are aware of theze possibilities and start to incorporate these ideas into their digital projects, Dizoussions with people in the field have shown that there is still come confusion about the role targets play in the digital imaging process, It is important to emphasize that targets are about the scanning system and not about collections, This rneans that the target evaluations are primarily aimed at characterizing scanning eysterns. At thie time, many aspects of ecarming photographs:me Fegure 11. The three tmages shove different repro~ duction ofthe gray tones. The image on the lft && too light. The one on the right has no details sn the shadovo areas and és too dark overall. The ‘one in the middie shows the most acceptable tone reproduction. National Archives and Revords Administration Monitor Adjustment Figure 12, Tone reproduction control targets for users looking at images on monitors are fmpor- tant quality control tools. This sample shows a gray wedge that is avatlable on the National Archives web site (www archives gov/research) arc/adjust-monitorhiml). The user ts asked to change brightness ard contrast on the monitor ‘ereen tn order to se all stepe on the gray wedge target, It is important to remember that images may have to be processed for wewtng-on a moni- tor. still require the intervention of a well-trained operator Ina few years, some of these tasks will he automated, and manual interventions will he lese and less necessary. One could say that targets will then he about collections, because no thatter what original ie scanned, it will auto- matically turn out right. However, targets will alwaye he ‘useful for checking the reproduction quality of the digital files, ¢.g., for confirming that aim-point values are actually reached. This does not mean that scanning is done entirely “by the numbers,” because an operator will till be needed to decide when to consciouely intervene to improve the subjective quality of the image. This doee not apply if 16- bit data are stored. ‘THE PARAMETERS OF THE IMAGE QUALITY FRAMEWORK Tone Reproduction ‘Tone reproduction is the single most important parameter for determining the quality of an image. If the tone repro- duction of an image is right, users will generally find the image acceptable, even if some of the other parameters are not optimal (Figures 11 and 12) ‘Tone reproduction is applicable only in the context of capture and display. This means that an assumption must he made regarding the final viewing device. Three mutually dependent attributes affect tone reproduction the opto-electronic conversion function (OECF), dynamic range, and flare, OECF ean be controlled to a certain extent via the scanning coftware but is alzo dependent on the A/D (analog to digital) converter of the scanning sys- tem; dynamic range and flare are inherent in the scanner hharchware iteelf ‘The OECF shows the relationship hetween the opti- cal densities of an original and the corresponding digital values of the file, It is the equivalent of the D-log H curve in conventional photography (Figure 13). Dynamic range refers to the capacity of the scanner to capture extreme density variations, The dynamic range of the scanner should meet or exceed the dynamic range of the original Flare is generated hy stray light in an optical system. Flare reduces the dynamic range of a scanner. ‘The most widely used values for bit-depth equivalen- cy of digital images ie still eight bits per pixel for mono- chrome images and 24 bite for color images. These values Building the Image Quality Framework255 are reasonably accurate for good-quality image output, Eight bits per channel on the input side is not sufficient for good-quality scanning of diverse originals. To accommo- Digital date all kinds of originals with different dynamic ranges, Value the initial quantization on the CCD side ranst be langer than eight bits, CCDs work linearly to intensity (transmittance or re- 0 flectance). To scan images having a lange dynamic range, o 1 2 83 4 12 to 14 real hits (hits without noise) are necessary on the Doneity input side, If theee hits are available to the user and can be aved, itis referred to as having access to the rare acan Figure 12. Using a calibrated gray scale target High-bit-depth information ie very important, espe- and the resulting digital values, the tone repro cially in the case of negatives. Negatives can be considered duction of the scanning deviee ean be determined. a photographic intermediate; they are not yet finalized for The reflection or tranamission density ofeach an “eye-pleasing” image like prints and slides. Negatives step of the gray scale can be measured with a ehow a high variability. They can have low contrast, high densitometer. Plotting these values against the contract, and everything in between, The dark parte of the digital values ofthe steps tn the image fle show the performance of the scanning device over the hole range of densities. negatives contain the important image information. Only very good scanners can resolve very well in these dark parts of the originale. The seanner needs a high dynamic range and not a lot of flare to produce real, noise-free, high-bit information. Often, it is only poseible to get eight-bit data out of the scanner The higher-bit file is reduced internally. This can be done in different ways. The scanner OECF shows how this is done for a specifie seanner at specific settings. It is often done nonlinearly (nonlinear in intensity, but linear in lightness or brightness, or density), using perceptually compact encoding. A distribution of the tones linear to the density of the original leaves headroom for further provess- ing, hut, unless the images and the output device have the same contrast, the images will need to be processed before viewing, Processing images to look good on a monitor will limit certain processing possibilities in the future. ‘The data reeulting from the evaluation of the OECF target is the basis for all subsequent image quality param- eter evaluations, e.g, resolution, Itis therefore very impor- tant that this evaluation is done carefully. In eases where data is reduced to eight bits, the OEGF data provide a map for linearizing the data to intensity (transmittance or reflectance) by applying the reverse OECF function. This Fieure 14. Calibrated gray-scole test targets erce _eP needed to calculate all the other parameters, Inthe to link back tothe realty ofthe orginal dora. c22e 0F 16-bit data, Iinearity to transmittance and reflec: iment or photograph tance will be checked with the OECF data, Any procesting 18 Digital Imaging for Photographic Collections: Foundations for Technical Standards0 o Geny lev 255 Figure 15, Histograms of the image files can be used to check whether all dittal levels from 0 to 255 are used (A), whether any clipping (loss of shadow and/or highlight details) occurred during scanning (B), or whether the digttal values are unevenly distributed as can be the ease after im- ‘age manipulation (C). Figure 16, Targets for measuring linearity of the ‘scanning aystem. The target om the Lover left ia for use with digttal cameras, and the one an the ‘upper right és for use with live acanners to linearize the data to density will fallow later: Repoducing the gray scale correctly often dacs not result in optimal reproduction of the images. However, the gray scale can be used ae a trail marker for the protection, of an institution’ inveetrment in the digital scans (Figure 14); having a calibrated gray eoale aesoeiated with the image makes it possible ta go back to the original stage after transformations, and it also facilitates the creation of derivatives. The gray scale omuld he part of the image, or the file header eould contain the digital values ‘Tone and color corrections on eight-hit images chould, be avoided. Such corectione cause the existing levels to be compressed even farther, no matter what kind of operation is executed. Ta avoid the lose of additional brightness reso- Iution, all necessary image processing should be done on a hhigher-hit-depth file, Requantization to eight-hit imasee should occur after any tone and eclor corrections Often, benchmark values for the endpoints of the RGB levels are specified by the institution. The National Archives, for example, ask in their guidelines for RCB levels ranging from & to 247.” The dynamic headroom at both ends of the ecale is ta ensure no logs of detail or lip- ping in ecanning and to accommodate the slight expan- sion of the tonal range duo to charpening or other image processing steps (Figure 15). Additionally, as part of the tone reproduction test, the flare of the system can be tested. Flare exists in every opti- cal eyetem, reducing the contrast of the original Targets to Use * OECF target for measuring linearity.” This target (Figure 16) characterizes the relation- hip between the input values and the digital output values of the scanning eystern. It is used to determine and change the tone reproduction. ‘The target was developed based on the ongo- ing research of the Electronic Still Photography Standards Group (SO/TC 42WC18), The target has heen manufactured under IPT’ guidanee by company in Rochester Since the specifications in the standard are tight, the production process proved ta be lengthy. * Flare measurement target. A flare model ean be determined with the various tone reproduction Building the Image Quality FrameworFigure 17, Tone reproduction targets can be used to build a flare model for the scanning system. This requires targets with diferent contrast and different backgrounds, targets. Targets with different backgrounds and. different dynamic ranges of the gray patches had to be manufactured (Figure 17).* Another approach consists of using a target with anarea of D,,, and an area of D,,,. allowing the measurement of the flare, ie., showing how much the original contract is reduced in the digital file (cee Figure 28 on page 24), In addition, this target can he used to check the registration of the three color channels for color scans, In case of misregistration, color artifacts wall appear at the edges. They can aleo be calculated (cee ref 34, Annex C). Detail and Edge Reproduction (Resolution) Monitoring digital projects showed that people are most concerned ahout spatial resolution issues, This is not surprising, because, of all the weak links in digital capture, spatial resolution has heen the best understood by most people, Technology has evolved, however, and today “rea- sonable” epatial resolution is neither extremely expensive nor does it cost a lot to store the large data files. Beeanee «questions concerning spatial resolution came up so often, we looked at this issue very closely. Spatial resolution is either input- or output-oriented. In the former case, the gal is to capture all the information that is in the original photograph; in the latter cage, the earning resotution is chosen according to fc desired output. Spatial resolution is the parameter to define detail and edge reproduction. Spatial resalution of a digital im- age, i., the number of details an image contains, is usu- ally defined by the number of pixels par inch (ppi). Spatial resolution of output devicee, such as monitors or printers, is usually given in dots per inch (dpi). To find the equivalent number of pixels that describe the information content of a specific photographic enaul- sion is not a straightforward process. Format of the orisi- nal, film grain, film resolution, resolution of the camera lens, f-stop, lighting conditions, focus, blur, and processing have to be taken into consideration to accurately deter- mine the actual information content of a specific picture ‘The following table gives an idea of the pixel equivalencies for-various film types.A B fa ® ep, COO ro) ® Figare 18, The number of sampling points within ix ginen distance ts refered bo ax the denice’s digital resolution. In A(2), the digital resolution 2 fous, and not ail the mage information will be included in the digial fle. However, scanners often do not scan correctly at tow resolution, showing the behanior deptised tn AD. Since the (piel values are taken over x cartain distemce and not averaged oder a certain area, the quality of he dow- resolution image in A(3) will be tower than in A(t). B(2) t9 sampled with a higher reso~ lation. A low-resolution derivative, B(2), that és calctdated from shis high-resolution fle (using a rescoapling algorithre) will ane a higher quality than the tmage originally scanned at low resolt- tion with method A(2) ‘Sampling Rasolatton for Batrastion of the Flim Resolution Flim spsed very low (< 64 18Q); 9500-5000 pizets{inch Rim spead madiurn (200-820). 2000-2800 pizsls/inch Sous [sk Hg FB Resa Las, Tis aan ofipal Phonan Seto She set Nu [84T 4 faa Case, ay 8 Another common question that should be answered before digitization starts ig, given the spatial resolution of the files, how big an ourput is possible from the available file size? The relationship between the size of a digital image file, it total nurnber of pixele, and coneequently ite maximum eutput size at different spatial resclutione ean be analyzed mathematically. ‘To answer the question, the distinction has to be made between continuous-tone and halftone output. For optimal continucus-tone output the ratio between output dote and image pixels chould be 1:1. In the eaee of print- ing processes that require halftone images, hetween 15:1 and 2:1 oversampling (pixels per inch of the digital file is one and a half to two times greater than dots per inch of the output) is needed. ‘What is Digital Resolution? Why do we measure resolution? We do eo, fret, to make sure that the information eontent of the original image is represented in the digital image and, eacond, to eneure that the scanning unit used to digitize the image is in fo- us. Unlike photographie resolution, digital recclution does not depend on vieual detection or observation of an image. Digital resolution is calculated directly frona the physical center-to-conter epacing between each eample or dot. This spacing ie algo called the sampling interoal (Figute 18) ‘The number of eampling points within a given dis- tance (usually an inch) is referred to as the device's digital resolution, given in ppi. Digital resclution quantifies the number of sampling dots per unit distance while photo graphic reeolution quantifies obgerved feature paire per unit dietanoe, as in line paire per millimeter (Ip/mmn} ‘Translating between units of classical reeolution and digital resolution is simply a.matter of “two.” Dividing Aigital resolution values in half will yield unite that are equivalent to photographic resolution, But there are con- ceptual differenoss between the two that have to be kept in mind when using digital rasalution. A mieregistration Quality Framework 21HEE EEE ttttt ttttt o ws 12 10 128 128 Dial Vee Dita ato Figure 19, Misregistration between the detectors and the lines of the target by half « pével can lead to the situation where the black-and-white lines ofthe original cannot be resolved and wild look like a gray field tn the digital image (all the digital values are the same, ée., 128) ‘Undersampled image line Figure 20, Aliasing occurs when an image is insufficiently sampled. The samples ell show up as a lower frequency: Example: An original has 500 line patrs per inch. Ths represents finetéon wtth 500 eyeles per inch. To capture tall the information (ald the valleys and peaks of the function), at least 1000 péxela per inch are necessary, this is the so-called “sampling rate” (1000/25.4=40 pixels per mm), The maximum Frequency a sampling device ean capture ts ¥> the sampling rate. This ts called the Nyquist frequency: In practice, more pixels per inch are necessary to capture all the information faithfully: In this sample the 1000-ppi deviee has a Nyquist Frequency of 500 line patrs per inch. 22 Digital Imaging for Photographic Collection: between image details and image sensors may give the impression that a certain device has les eealution than it actually has. Furthermore, aliasing is an iseue. In an ideal eean, the detectors and the lines of the target are perfectly aligned. The concept of misregistra~ tion ean he easily shown by scanning a bar target. "The detectore-will only sense the light intensity of either the black line or the white space. If there is a misregistra- tion between the centers of the lines and spaces relative to the detector centers, say by half a pixel, the outcome is, different (Figure 19). Now each detector “sees” hall a line and half a space. Since the output of every detector is just 1 single value, the intensities of the line and the space are averaged. The resulting image will therefore have the same digital value in every pixel. In other words, it will look lke a gray field. The target would not be resolved "Therefore, the micregistration manifests itself as a contrast or signal, lose in the digital image that affects resolution, Since it is impossible to predict whether a document's features will align perfectly with the fixed positions of a scanners detec- tors, more than two samples per line pair are required for reliable information seanning. If the sampling interval is fine enough to locate the peaks and valleys of any given sine wave, then that fre- quency component can he unambiguously reconstructed from ite sampled values. This is referred to as Nyquist fre~ quency. Aliasing occurs when a wave form is insufficiently sampled (Figure 20). If the sampling is lee frecquent, then the samples will be seen as representing a lower-frequency ‘The most noticeable artifact of aliasing is high spatial frequencies appearing as low spatial frequencies. After the wave form has heen sampled, aliasing cannot he removed by filtering. How Is Digital Resolution Measured? ‘The fundamental method for measuring resolution is to capture an image of a suitable test chart with the scanner being tested. The test chart must include patterns with sufficiently fine detail, euch as edges, lines, square waves, or sine-wave patterns ‘The Modulation Transfer Funetion (MTF) ‘The best overall measure of detail and resolution is the modulation transfer function or MTR MTF was developed to deseribe image quality in elassieal optical cystems—eo-Figure 21. Graph of the modulation transfer function. The MTF shows the performance of two scanning aystems over the whole range of {frequencies (inthe target represented by sine ‘waves that are elaser and closer together). The cutoff point of the aystem represents the high cat frequeney (the finest image detaila) that the scarier to able to resolve. An MTF speesfcation to be met by vendors must be established. Scanner A has a better performance than Scanner B; scans _from Scanner A will appear sharper 1 MIF Nrguist Frequency v ie 1 Frequency, cy/mm 10 Figure 22, Graph of the modulation transfer fane- tion for comparison of different imaging systems. I the tested system shows MTF values above the straight line (area with patter), it performs well enough. The numbers tn this figure apply for ‘scanning at 500 ppl. called linear systems, The MTF is a graphical representa- tion of image quality that eliminates the need for decision- making by the observer. The test objects are sine-wave patterns If the MTF is measured for a sarnplec-data systera that is nonlinear, we are talking about the spatial frequency response (SFR) of the system, The terms MTF and SFR are ured interchangeably. The measured results for a sampled- lata system will depend on the alignment of the target and. the sampling sites. An average MTF ean he defined, azeun- ing that the scene being imaged is ranclornly positioned with respect to the earnpling cites Output Modtitation MIP at Modiation — (across a range of frequencies) ‘The MTF is a graph that represents the image contrast relative to the abject contrast on the vertical axis over the range of spatial frequencies on the horizontal axis, where high frecqueney in the test target corresponds to eral detail in an object (Figures 21 and 22), ‘The MTF should be cletermined and reported for both horizontal and vertical sean directions, sines the results can differ MTF is difficult and cumbersome ta measure in images on photographie materials, requiring epecial equipment, like a microdensitometer Tris relatively easy to measure in, digital images, however ‘There are two approaches to defining the MTF of an imaging eystem. One is ta use a sine-wave pattern (Figures 23-26), the other is to use a slanted edge (Figures 27 and 28), In the latter ease, pixel values near elanted vertical and horizontal black-to-white edges are digitized and used 7mm, Spatial frequency = cyclo Armpit Figure 29. A wave is characterized by its amplitude and its spatial frequency.‘Test Target ° ‘Capui) 1255, Pixel Values in Digital Image (Output) Figure 24, The sine waves of the test target are scanned and transtated into digital calues. Ifyou twere to measure hot dark or light the image is at every point along a line across the bars, the plot of these points would be a perfect sine wave. Figure 25, Input modulation/output modulation DW vnc + + + NAMA AMG Ih AMV Wes GN Figure 26, Aa the bare of the eine-toave target get closer together at higher frequencies, the modula tion (i.., variation from black to white) that fs recorded by the scanner gets smadler and smaller 24 Digital Imaging for Photographic Collection: =. “LZ Figure 27. Calculating the MTF using the moving knife-edee ‘method. Péxel values across a slanted edge are digitized and, through a mathematical transformation of these calues into the Fourier domain, the MTF of the system can be calculated. to compute the MTF values The use of a slanted edge allows the edge gradient to be measured at many phases relative to the image sensor clemente, in order to eliminate the effects of aliasing, (This technique is mathematically equivalent to performing a “maving kniffe-edge measure- ment.”) Figure 28, Krafe-edge tar- get for resolution measure~ ment, Car also be used for flare measurement. Targets to Use At this time, the bar targets designed for measurement of photographic resalution often are used to measure digital ‘resolution (Figure 29). These targets are nat suitable for this task, except for visually checking aliasing. Visually checking har targets is mot an eacy tack; the observer must know what to look for. Therefore, to measure digital resolution of sampling deviove another ajsproach has to be taken using slanted edges or sine-wave patterns * Knife-edge target for resolution measure ment. A knife-edge target has hean developed by the Electronic Still Photography Standards Group (Figure 30). A special target (QA-62) for testing scanners is available now as well. We used another target (Figure 28) that looks like the one used for flare measurements. It has been thoroughly tested using a specially developed enftware module. "Tests showed that the target in its eurrent form, is usable for a ean resolution of up to 600 dpi.Figure 29. Bar targets with converging tines, like this star patter, cam be used to visuadly check the so-called cutoff Frequency of the aystem (ie, the small- cst features that can be resolved), but they cannot be used to get information on how the gyster és working for al the different Frequencies inthe image. Figure 30, Resolution test chart developed for els tronic still photography: The black bars are used to calculate the modulation transfer farction. TT Figure 31, Sine Pattes sine-wave target. The sine waves in the treo center rows of the image are used to calculate the MTF. Figure 32, Software fnterface to compare MIF calculations from sine-wave target and krafe-edge tare igure 33. User interface of the software tool fo ‘measure MTP from a Inifecedge target devel- oped at IPI. This i cue to limitations in the in-house produe- tion process. Plane for developing a test target for higher scanning resolution were abandoned. + Sine-wave target by Sine Patterns Ine. for resolution measurements. This target (Fig- ure 31) exists on photographie paper and film in different formats, For analysis of the sine-wave target, software developed by MITRE was used." After testing and comparing the sine-pattern and Iknife-edge techniques for determining MTF, IPI decicled to ‘nclude both mathods in the test (Figures 32 and 83). The remults indicated that both methods produce similar MTF characterizations under proper scanning conditions. It was found that the two methods recponded differently to image sharpening, nonlinear output gamma mapping, anc im- proper resampling methods, The advantage of using both methods simultaneously lies in the ability to detect unde-Dengity Figure 24, Moise vefe's to vandom signe foes. tions in elartonie imaging systums, represented as.arandom pattern in the uniform gray pob- tans. This graph shows the RMS (root meare square) evr, «statis teal evadua.tion representing aise over the density els of the original image. An RUB of Dis egunl to 10 noise. arable post-scan processing that has been intentionally or unintentionally applied The software module developed. at IPlto calculate the MTF for a knife edge can also compare the sine-wave and laife-edge techniques, Noise 4s defined estlier, noise refers to random signel variations associated with detection and reproduction systems. In conventional photography, noise in an innage is the graini. ness that can ke perceived; it ean he seen most easly in uniforrn density areas. The objective measure is granulari- ty In electronic imaging, noise is the presence of unwanted energy in the image signal (Figure 34) Noise is an important attribute of electronic im. aging systems * Standardization wil assist users and maantfachurersin determining the quality af images heing produced by these systems. The visihility of noise to human okservers depends on the magnitude of the noise, the apparent tone of the area containing the noise, the type of noise, and the noise frequen cy. The magnitude of the noise present in an output representation depends on the noise present in the stored image data, the contrast amplification or gain applied to the data in processing, and ‘the noise inherentin the output pro cess and media, Noise visibility is different for the luminance (monochrome) channel and the color channels ‘The result of the noise test is twofcld. Fist, it shows the noise level of the system, indi cating how many bit levels of the image data are actually useful. For example, if the specifications for a scanner state that 16 hits per chamnel are reconed cn the input side, itis impertant ta low how many of these hits are actual image informaa- tim and how many are nize; in most cases, two hits are noise. Far image quality considerations, the signal-to-noise ratio (S/N) is the important fa ctor to mow. The noise of the hantware wed should not change unless the scanner operator changes the way she/he works or dirt is built up in the aystern ‘There exist diffrent types of noise in an imaging sys- tem. The following definitions are part of the forthcoming standanl for noise measurement, + Total noise. all the unw anted variations captured hoy a single exposure (scan)Figure 35, Target used for noise measurements (bce BO Test Charts, p. #8) Figure 86, Matching original, soft copy on the monitor, and hard copy is she goal of color management + Fixed pattern noise. The unwanted variations that are consistent for every expoeure. + Temporally varying noise. Random noise due to senor dark current, photon shot noise, analogue processing, and quantization that varies, from one image to the next, All of these parameters should be measured and the three above-described signal-to-noise ratios reported for the imaging system. Since many electronic imaging systems use extensive image proceecing to reduce the noise in uniform areas, the noice measured in the different large area gray patches may not he representative of the noise levels found in scans from real sosnes. Therefore, another form of noise, so-called ze noise, will have tobe looked at more dovely in the future Target to Use ‘The OECF target shown in Figure 16 that was manufac- tured to measure tone reproduction ean be used, under certain circumstances, to measure noize over a wide range of input values. A epecial noize target (Figure 35) that lao includes patches with frequency patterns has been devel- oped by ANSI IT-10. Color Reproduction In addition to the image quality framework on the input side, color reproduction and tone reproduction in all areas of the image databace were looked ar more eleeely in the second phace of the project. This report will not go into detail in this area. The following chows a theoretical ap- proach that IPI has been developing over recent months. Because ofits complexity, color reproduction (Figure 36} wae heyond the eoape ofthis project; there aro, how- ever a few thoughts to he considered 5 ‘Two basic types of image databace can he defined, each needing different approaches to creation, access, and use of images. The fire type of image database is the one containing large numbers of images, e.g, an image database for an insurance company, a goverment agency, or a historical society. In this caee, the intentiis to have in the database a good-quality image that is not neoossarily colorimetrically ruc to the original. Its the information content of the image that is important, It is agreed upon Building the Imaze Quality Framework 27sgraphic Collection from the beginning that the digital file in the image da- tabase will not be used to recansteuct the original phota- graph or soene. ‘The other type of image databaze will he built in en- vironments where it is important that the digital image is “replacement” of the original. Therefore, the digital data has to be captured and archived in such a manner that the original can be reconstructed Pictorial Rendering Intent When tangeting a wide user audience, computer platforms, colar management systems, calibration procechures, color space conversions, ar output devices cannat he mandated Almoet every institution hae a different cetup for acceze of the images, In addition, at the time of creation, it is ueually nat known what type of hardware and software are or will be available for access. Furthermore, one has to consider that the usere of image databases are usually not trained in matters of digital imaging. However, they are visually sophisticated and are used to the quality they can get from conventional photagraphic images Nevertheless, decisions have to he made about spatial resolution, tone reproduction, and color space before im ages are digitized. In most eases, it will not be the goal to reproduce the physical properties of the original, but to reproduce its appearance under certain viewing conditions, Ascumptions about the rendering device, color reproduc- tion, and tone reproduction have ta be made. All original artwork cannot be treated! equally. Differ- ent digitizing approaches have to he established according, to the type, condition, and perceived values of the originals. Current scanning technology deals reasonably well with current film emulsions and formats, provicled that they have been exposed and processed correctly. However, many collections of high artistic and/or historical value were eap- tured on photographie material that not only isn't available anymore, but also has deteriorated to some degree. ‘There are several rendering intents that apply while digitizing original artwork * The photographic image is rendered. In this case, the images are scanned with the intent to match the appearance of the original photographic image. The quality of the digital image can be evaluated by visually comparing the original to a reproduction on a ealihrated display device withFigure 37, Matching the appearance of the origi= nal photograph will be the prime goad in many archives databases. Figure 38, An undereeposed slide might be car- rreted during scanning to render the photogra- pher’s intent a similar contrast range. The assumption is made that the original photograph has heen exposed and proceseed perfectly (Figure 37). The photographer's intent is rendered. There are many photographs with high content value that were not expased or processed comectly. They can have a color cast, be over- or underexposed, or have the wrong contract. In these caees, the photographer's intent, nat the original photo- graph, needs to he rendered to achieve a pleasing reproduction. The soanner operator hasta raake decisions about tone and color reproduction by viewing the digitized image on a calibrated output, device, This manual intervention determaines the quality of the reproduction, Quality becomes highly dependent on the skill and experience of the opera- tor (Figure 38), ‘The original appearance of the photograph is rendered. Often, older color photagraphe are faded and no longer have sufficient visual color information to make accurate judgments about the original, Reconstructing these photographs requires special esanning and processing techniqques. + The original scene is rendered. When pho- tographio reproductions of original artwork are canned, the original scene has to be rendered and the film charactenstics have to be subtracted. However, this is only possible if a target is included on the film and the film characteristics and lighting, conditions are known, It is therefore probably bet- ter to scan the original artwork With current scanning and color management technol ogy, the first cae, rendering the photographie image, ean hhe automated ifit is possible to match the dynamic range of the output to the original. All other cases need manual intervention, either in the initial scanning process or in subsequent image processing. Manual intervention is time- consuming and recquires highly killed operators. As a result, produotion coets remain high for high-quality, vieually pleasing digital images. Better automated image processing tools need to be developed to analyze raw sensor data andl translate it to pictorially pleasing digital reproductions on. defined output devices,INTERNATIONAL COLOR CONSORTIUM (CC) The ICC was established for the purpose of creating, promoting, and encouraging the standard. ization and evolution of an open, vendor-neutral, cross-platform color management gystem. ICC standards will be important for the color fidelity of image da- tabases, since ICC profiles are being used in some projects in the field. Currently, the ICC defines @ profile format that contains ‘mapping: information from input or output device color space to a PCS (profile connection space) How this mapping is achieved is vendor-specifc. As a result, mix- ing profiles from different vendlora (for the same devices can result in different image reproduction. However, using profiles is only a temporary solution for archives. In the long term, other solutions aareneeded that are more open and that do not include any propri- etary technology: Digital Imaging for Photographic Collection Choosing a Color Space ‘Tho most important attribute of a color space in an archi val environment is that itis well defined. Scanning for an image archive is different from scanning for commercial offset printing. When an image is scanned for archival purpoees, the future use of the imago is not known, nor are the technical options that will be available a few years from now. Will color profile still be maintained er even used? An eight-hit-per-color scanning device output might be sufficient for visual representation on today’s output devices, but it might not capture all the tonal subtleties of the original. Operator judaments regarding color and con- trast cannot he reversed in a 24-bit RCB color system. Any output mapping diffrent from the archived image’s color space and gamma must be considered. On the other hand, saving raw scamner data of 12 or 16 bits per color with no tonal mapping can create problems for future output if the scanner characteristics are not well known and profiled. As an option, a transformation can he associated with the raw scanner data to define the pictorial intent that was chosen at the time of capture, However, there is currently no gofiware available that allows one to define the render- ing intent of the image in the scanner profile, Rendering intent ie usually set during output mapping hy the user ‘Thore is coftware available that allows the user to modify the ccamner profile for indiviclual images, and therefore to create “image profiles.” That procese is as work-intensive as regular image editing with the seanner or image pro- cessing coftware. It also assumes that the input profiles ean and-will he read by the operating eystem and application the image is used in, not just by current hut also by fusture systems Archiving for each image both a raw sensor data file in high bit-depth and a calibrated RGB 24-bit file at high resolution is not an option for a lot of institutions, consid ering the number of digital images an archive can contain Because the output is not known at the time of ar- chiving, it is hest to stay as clase as possible to the source, ive, the scanning device. In addition, scanning devices bould be spectrally well characterized, and the informaa~ tion should be readily available from the manufacturers, New Tools and Developments Archiving 10- to 12-bit-per-channel standardized RGB color space would be optimal. Having to communicateThe ultimate goal iva truly seam- less, transparent work flow of images across platforms, devices, and, ultimately; time. only one color space (ar profile) to the user's calor man- agement system would facilitate optimal rendering of all images across all platforme and devices. Ifthe color space wore standardized and universally recognized, the need to embed a profile into each image file would be eliminated. (Erabeddling profiles into each image file creates too great adata averhead when delivering preview files aver the In- temet.} There would alsa be only one profile that needs to be updated when calor management specifications evolve in the future ‘The eRCB color space propoeed by Hewlett-Packard and Mieroeoft (or an extended version allowing the ac- comamedation of an unlimited garaut and out-af- garsut calars} is a viable color space choice for access files. i euficiently lange to accammedate mast photographic reproduction intents. Since the first future access to any. file will most probably he some kind of a monitor using an RGB colar epace, choosing to keep the accese data in the currently defined sRCB is a valid solution. Images in ROB will display reasonably well even on uncalibrated moni- tors. Higher bit-depth per channel would make it possible to communicate the predefined rendering intent for each image while leaving enough bit-depth for users to modify the image and to map to the intended output device. It would also give a safety factor to the archive fle if future high-quality output devices require extensive mapping to as yet unknown color garnut and gamma. Also, colors that currently fall out of gamut could ati be accounted for by leaving enough room on both ends of the values scale when defining black and white values. A standard way to deal with higher than eight-bit-per-channel image data across platforms and applications has to he developed It has heen encouraging to see the development of high-quality tools destined tor digital image applications other than prepress. However, there is still a need for mare integrated systems to achieve a truly seamless, transpar- ent work flow of images acroes platforme, devices, and, ultimately, time, No one imaging technology manufacturer will ever he able to dictate to the end-user which imag ing system to use. Additional standards will have to he developed to facilitate communication between imaging systems and to enable high-quality digital imaging for im- age database appliostions Building the Image Quality FrameworMONITOR VIEWING CONDITIONS 180 3664 Viaving Conditions—for Graphic Technology and Photogra- ply) requires the following, * The chromaticity of the white displayed on the monitor should approvimate that of D65. The lu- rminance level of the white displayed ‘on the monitor shall be greater than 7 edim2 and should be greater than 100 clin? + Phen measured in any plane around the monitor or observer, the level of ambient illumination shall be less thar 64 lice ond should be less thane 82 lux. The color temperature ofthe ambient illumination shall be less than or equat to that of the monitor white point * The area immediately surround ing, the displayed image shall be neutral, preferably grey or black to minimize flare, and of appraxi- mately the same chromaticity as the white point of the monitor. + The monitor shall be situated $0 there are no strongly colored areas (including clothing) directly in the field of view or which may cause reflections in the monitor screen Ideally atl walls, floors, and fur- niture in the field of view should be grey and free of any posters, notices, pictures, wording, or any other object which may affect the *# Allsources of wlare should be avoided since they significantly degrade the quality of the image. The monitor shall be situated s0 that ro illami- ration sources such as unshielded lamps or windows are directly in the field of view or are causing reflections from the surface of the monitor ital Imaging for Photographic Collection Image Artifacts In addition to the four parameters described above, itis, important to check for image artifacts such as drop-out lines, banding, ete. These artifacts can be consistent from, image to image. There is not much that can be done about these types of artifacts, since they are introduced by the zeneor or the network connection hetween the sensor and the CPU. Serine Up Imacine Systems Acommon problem when using different computer systems or monitors in an environment is the difference between the images when viewed on the various systems. Systeme need to be set up and calibrated earefully. More often than not thie is not done properly, leading to various problems. For example, even if syetems are actually being: calibrated, measurements may not be taken correctly. Monitor Calibration In many digitization projects monitor calibration is an im- portant consideration, not only when working with images but also when discuseing the quality of scans with vendors over the telephone. IF monitors are not properly calibrated, the two parties will not see the same image. To solve this potential problem, the National Archives, for example, have the same system setup for their quality control station as their vendor If the monitor is the defined output device, ast is for many projects, it needs to be ealibrated to a specifie white point and gamma, A monitor’s gamma is a measure of the response curve of each of the red, green, and blue chan- nels, from black to full intensity. Typical gamma values for color monitors range from 1.8 to 2.2. For PC systems the latter value is usually chosen; in Mac environments: 1.8is the gamma value that ie widely used, The white point of a monitor is the color of white produced when all three color channels are at full intensity. It is epecified as @ color temperature measured in Kelvin (with images getting bluer ae their color temperatures rize). Typioal for this type of application is a setting to 6500° Kelvin. ‘There exist various calibration methods and tools that differ widely in coraplexity Some application programs incorporate hasie monitor calibration. Furthermore, thereAFEW POINTS TO REMEMBER * The area used for scanning needs tobe big enough to accommodate Preparation of: Images Scanning Quality control. *+ Critical points to be checked after scanning: Sharpness Correct file name Lateraily reversed images exist specific calibration programs, Depending on the need of the user, they can be very sophisticated and incorporate devives like photometers and colorimeters, ‘The best way to view a monitor is under dira illurni- nation that has « lower correlated color temperature than, the monitor, This reduces veiling glare, increases the rnoni- tor dynamic range, and enables the human visual system to adapt to the monitor. Thie viewing condition results in ‘the most aesthetically pleasing monitor images. Viewing, originals and images on the screen side-by-side is more problematic, because in this case the observers are not al- lowed to adapt to each “environment” individually, Once calibrated, the monitor should need re-calibra- tion only when conditions change, or on a monthly basis. It is a good idea to put a piece of tape aver the monitor’e brightness and contrast controls after calibration and to ‘maintain consistent lighting conditions Dierma Master AND DERIVATIVES It has heen agreed upon in the preservation community that eoveral files chould be stored for every image to fulfill all requirements, mainly preservation and access. First, a 20-oalled archive file containing more than eight hits per channel should be stared. It should nat be treated for any epecific output and chould be uncompressed or lossless compressed. From the archive file, various accece filee canbe produced az needed. These might be based on a particular use that defines tone reproduetion, and color reproduction, and pictorial interpretation. ‘The highest quality file produced is referred to as the digital master. The quality level of the digital master will depend on the goals of the project, and in most cases its level of quality will be dictated by the project bud- get, From this digital master several derivatives can be produced, Usually up to five different quality levels will be produced (see, for example, www.nara, gov). Again, this depends on the project. Quauiry anp Process ContROL. ‘The best approach to digital image quality control includes, on one hand, subjective visual inspection and, on the other hand, objective measurements performed in software on the digital files themselves. Efforts should he made tosgraphic Collection standandize the procedures and equipment far subjective evaluations by means of monitor and printer calibration. For objective image cquality measnremant, software should be available which is designed to locate and evaluate spe- cific targets and then report numbers or graphe describing key imag quality parameters, Such software should ideally be a plug-in to a full-featured image browser eo that all aspects of the image file (header info, index, and tracking data, etc.) can he reviewed at one time. Some software components already exist, others are currently being devel- oped. A key point is that targets and the eofeware to evah- ate them are nat just for checking systems—they serve to guarantee the long-term usefulness of the digital files and ‘ts protect the investments of the institution Since come of the targets and software described ere are not yet commercially available, it will till be come time before consietent quality control can be cot up. Neverthe- less, these tools will be available soon, and their inclusion in future work flow is recommended. Sample materiale representative of the photographs in a collection to be scanned chould always he included in the syetem tests Ibis crucial that eample materials are a true reprogentation of the collection. Taking thece carmple images through the whole processing chain will help to ascertain whether a particular quality level can be reached. Depend- ing on the goals of the digitization project, this process may inclucle working with outside vendors, printers, and the like. Benchmarking: Scanner Systems Standards are currently being developed for benchmarking, scanning systems. In most cases the test requirements will far exceed the necessary performance for actual scanning. Benchmarking systems will help to compare different hardl- ‘ware, give more adecuate information than that whieh ie currently available from the manufacturers, and hopefully lead to a better understancling of the whole process Reproduction Qualities and Characteristics of the Digital Master and Derivatives ‘The reproduction qualities of digital images will have to be monitored during and controlled after their production, Hence, a quality-control step needs ta be incorporated into every production step. This process eerves to ensure thatMETADATA: “DATA ABOUT THE DATA’ Discovery metadata for finding. Adininistrative metadata for view- ing and monitoring. Structural metadata for naviga- tion, Rights-management metadata for controlling access. images actually reach the quality margins that have been set for the various parameters. Funetional Qualities and Characteristies of the Digital Master und Derivatives Besicles tests for reproduction quality, additional tests are needed to make sure that the fimetionality of the digital master is inherent in the digital file. These include looking at file formats and performance, for example, always keep- ing in mind the longevity of the data. IPI recommends the use of standard file formats each as TIFF, which is thoroughly published. This ensures independence from proprietary formats that might one day be digcontinued. New file formats like TIFF/EP® or eomething similar, might be more widely ueed in the future ‘TIFF/EP is hased on TIEF 6.0 and uses a lange mumber of tage to store additional information, JPEG2000 is another standard that should be considered Documentation of the Imaging Process—Technical Metadata (Subcategory of Administrative Metadata} ‘To be able to work with images across platforms, as well as over time, itis important that the imaging process is well documented and that the information is kept with every file, The National Information Standards Organization has published a standard called Technical Metadata for Digital Stil inages.® Tt represents a. comprehensive list of techni- cal metadata elements recqured to manage digital image collections IMsce PRocessinc Image quality is affected by the sequence of applying different image processing steps." It is important ta be aware of the effects of diffsrent processing algorithms (e.2., resampling, sharpening and noise reduction, tone and color conection, defect elimination, and compression). It aleo has to be kept in mind that the future use of the images ie not clear. Ideally, all image processing should be delayed until the time an image is actually used and image rendering, and output characteristics are known. This would require the data to he stored with a bit-depth of more than eight bits per channel. Unfortunately, most work-flow eolutions currently available do not allow this. Building the Image Quality FrameworPROCESSING FOR MONITOR VIEWING, * A linear distribution of the tones in a digital image compared to the density values of the original offers greater potential for future functionality: but images need to be adjusted before being viewed on a monitor. + Adjusting master files for moni- tor representation provides better viewing fidelity but means giving Lup certain processing possibilities in the future. DATA COMPRESSION © Por archiving: uncompressed © Forse: lossy 1" Photographie Collection Processing for Archiving Image data is best etored az raw capture data. Subeequent processing of this data can only reduce the information content, and there is always the possibility that better input proceesing algorithme will become available further on. The archived image data should therefore be the raw data, and, when possible, the associated information required for processing, such as sensor characteristics, il- Iumination, and linearization data. Processing, for Access Processing for viewing is a type of output processing applied to produce images of good viewing quality. It is possible to design viewer software that can take image files that have undergone input processing and process them, for output on a monitor: Data Compression Advances in image-data compression and storage-media development have helped to reduce the concern about storage space for large data files. Nevertheless, image com- pression in an archival environment has to he evaluated very carefully. Because in this case the future use of a cigi- tal image ie not yet determined, one copy of every image ehould be left uncorapressed. Current loeslese compression, echemnes do not bring too rnuch in termas of reduction of storage space. Also, it should he remembered that the loss ofome crucial bit could mean the loss of all of the file infor- ‘mation, even in the case of lossless compression, New compression echemes, like wavelets, which do not produce the well-known artifacts that JPEG com- preceed files chow, are atill not readily available ‘Two of the most widely used compression echemes are briefly described below. Lossless and Viaually Loveless Compression AA good compression schemes eacrifice the least informa- tion possible in achieving a reduced file size. Lossless com- pression makes it possible to exactly reproduce the original image file from a compressed file, Lossless compression differs from visually lossless compression (compreseion where the artifacts are not vieible). Although the human vienal eyetem provides guidelines for designing visuallyTRask Figure 99, Boon lightly compressed JPEG (41/5: can show some compression orbfacts ‘For compression, the image i divided into Fx 8 peel blocks; these blocks ove a mujor source of isible antifacts, Therefore, the highest-quality ‘master fle should be archived using o looatose compression algorithn Iceslees compression echemes, ulimately the viibility of compression artifacts depends on the output. + LZW (Lossless Compression). LZW (Lempa- Ziv Welch) is a type of entropy-basad enooding. It belongs to a. dass of lossless compression that is performed on a digital image file to produce a amaaller fle which nevertheless contains all the information of the original fle, Currently, the moet common echemes are thoee baced on Huffman encoding and the proprietary LZW compression (used for TIFF files in Adobe Photoshop). Unfortunately, the granular structure of film hhindere effective enteopy-based enooding. The filra ‘grain innposee a fine random neige pattern on the image that does not oomprese well. There ie cur- rently no effective loeeless solution to this problem. + Photo CD (Visually Lossless Compression). The Photo CD compression scheme utilizes both frequeney and color compression in an attempt to produce visually loseless compreseion. Lossy Compression JPEG stands for Joint, Photographic Expert. Group, which is the group responsible for the development of the compression approach named after it. JPEG is one type of loeey compression with a nunaber of user-selectable options igure 3%, The advantages of JPEG oompreseion are ite user electability to ensure visually lossless compression, high compression ratio, good computational efficiency, and good film grain suppression characteristics. Future devel- ‘opment proposed for the JPEG standard allow for tiling extensions, meaning that multiple-resolution versions of an image can be stored within the same file (imilar te the concept behind the Photo CD files). ‘The concer that repeated JPEG compression causes deterioration of image quality is valid.®' Consequently, all inmage processing should occur before the file is eom- reseed, and the image should only be saved onee using JPEC Building the Imaye Quality Framework 37Ae THE CONFERENCE ‘Tho outcome of the IPT study was prezented in a three- day conference at RIT in June, 1997, During the course of the poject we had the opportunity ta build a network with a wide varisty of individuals working in the field of digitization. The concept for the conference was to bring, together experts from different backgrounds and to try to determine whether a dialogue wae possible andl desired by thoge in the file, Putting together the program wae a very interesting procese, and we were pleasantly surpriced that all the speakers immediately agreed ta come. We took this a an indication that the chosen approach was appreciated in the field ‘The conference, “Digitizing Photographic Collec- tions Where Ave We Now? What Does The Future Hold?” ‘was held in the new CIMS Imilding on the RIT campus Over 120 people attended, a third af whom came from outside the US. The program was well received by beth, epeakere and attendees. The topics prezented during the three-day event are listed on page 39. Digital Imaging for Photographic Collections: Foundations for Technical StandardDIGITIZING PHOTOGRAPHIC COLLECTIONS Tere 7-9, 19 INTRODUCTION Digitizing Photographic Collections: the Big Picture—James Rally andl Franziska Frey, IPE Finding Possibilties—Charlas Kolb, NEH IMAGE QUALITY Introduction to Digital Image Quality and a Look at Took to Control It James Reilly and Franziska Frey IPE PROJECTS Digitally Archiving Glass Plate Negatives—Féenraa Wilson, University of Fnginia Adventures in Reformatting: A Hybrid Ap proach—Connuie McCabe, Photo Preseruation Services, ie Digital Archive Solution for the US Holocaust Museum—Holly Frey; Applied Graphic Technologies IMAGE INTERCHANGE AND MIGRATION—NEEDSAND TOOLS FlashPix Image File Format—Chris Hau, Kodak Determining Conversion Requirements for Special Collections Material—Arnne Kenney. Comeld University ‘The Muscum Educational Site Licensing Project and lesues of Image Interchange andl Metadata—Howard Besse, WO Berkeley ‘SHARING EXPERIENCES Imaging Systems for the Creation of a Large Digital Archive—Sabine Stsstrank, Corbis Corp. Modern Times at the Picture Factory—Carl Peischhauer and Phil Michel, Library of Congress ‘Technical Overview of the National Archives’ Electronic Acces Project—Steve Puglia, National Archives ‘THE TECHNOLOGY—A CLOSER LOOKAT SOME DETAILS Visible Human Images: Compression Techniques—Geenge Thoma, National Library of Medicine Scanning Technology: Now and the Futuro—Poa Nord, Xerox Corporation Digital Color Printing Systems—Frank Cost, RIT NATIONAL AND INTERNATIONAL TRENDS IN IMAGE DATABASE PROTECTS Filling the Frame: Selected Intemational Efforts in Photographic Digitization — Nancy Etkington, Research Libraries Growp Digital Image Collections: Issues and Practice—Michael Ester Luna Imaging Digitizing Collections of Cultural Materials for Digital Libraries: Some Experi- ences—Fred Mintzer IBM DIGITIZE TO PRESERVE Imaging Investments: How to Negotiate for Quality and How to Measure It—Stene Chapman, Harvard University Preservation in the Digital World—Paul Coneony, Yale Univer Mean Prsreation Pits While Dg Photogsph- i Collection —Paud Messier Boston Art Conservation IMAGE QUALITY Project Presentation: Images of African-Americans in 19- Century Illustrations—Anthony Troncale, New York Pubic Library. PANEL DISCUSSION ‘Technical Ieeues and Foundations for Standardssgraphic Collection fe CONCLUSIONS With the advent of digital imaging, new approaches to preservation strategies are needed for photographie eollec- tions in libraries and archives, The preservation camanma- nity must be involved in designing these strategies. The practices that are being developed and used influence the development of the field asa whole, A great deal of exper tice has been accumulated through various projects However, there still remain questions on “how-to” that have to be solved, the sooner the better. Since buelgets are limited, most of the scanning probably will be (and should be) dane only once. If project outcomes are not satisfactory, it will not only be disappointing for the participating team, it will aleo be considerably harder to get funding for a second project, We see more development needed in incorporating cjuality control inta a work flow, solving color issues, and creating the right metadata to al- low the files to he transferred to future systems The more deeply we hecame involved with image quality, the more we saw that the imaging process ean- not be viewed in a linear fashion. All the components are interdependent and therefore must he viewed as a whole. Nevertheless, understanding the components singly is the hbasis for understanding how they influence each other The many advantagee of the emerging digital to nologive for photographie collections are obvious, but there is still a long way to go. The technology is till young andl in constant flux. Most importantly, comarmunication among all participating parties must be improved. Due to the fast-changing and complex imaging technologies involved, collection managers need to work together with engineers andl imaging ecientiets, who often lack collection-related knowledge. Each side must he willing to learn the special needs of the other The conference that was organized within this project demonstrated this. This ie perhaps the project’s most encouraging outcome for the preservation commaunity Finally, it rmnet be remembered that imaging is about images, Visual eophistication is needed to suecesefully mnas- ter a digital imaging project—and this is clearly plentiful in the world of museums, archives, and Hbraries,Fig REFERENCES [2A-WGIS Digital Photography, www i8a.org/wal8, hana J. Holm, “Survey of Developing Blectronic Photography Standands,” Critical Reviews of Optical Science and ‘Technology, SPIE, CR61, 1996, pp. 120-152 K Donovan, “Anatomy of an Imaging Project,” Spectra, 23 (2), 1995, pp. 19-22. R. Gschwind, L, Rosenthaler, and F Frey, “The Use of Digital Imaging for the Preservation of Collections of Photographs and Motion Pictures,” Proceedings, ICOM 1° Triroiial Mecting, Ecinburgh, Scotland, September 1.6, 1996, pp. 8-15, I-L, Bigourdan and J. M, Rell, Erairorment and Enclosures in Film Preservation, Final Report to National Endowment for the Humanities, Grant #PS20802- 94, Image Permaneno> Insitute, Rochester Institute of ‘Technology, September 1997, JI. Rothenbong, “Ensuring the Longevity of Digital Doou- ments,” Scientific American, 272.(1), January 1995, pp. 4242. Preserving Disital Information, Report of the Task Force on Archiving of Digital Information (Washington, DC: ‘Commission on Preservation ancl Access, May 1996), Tine & Bits, Managing Digital Continaity, September 1998, wwvelongnow.org/projectsoonferencesitime-and- bitybackground, P McClung, ed, REGDiital bnage Access Project (Mountain View, CA: The Research Libraries Group, Ine, 1995) L.Serenzon Colet, K Keller, and E. Landabens, “Digi- tizing Photographic Collections: A Case Study at the ‘Museum of Modern Art, NY,” presented at the Electronic Imaging and the Visual Arts Conference, Paris, Septem- her 1997, S Ostrow, Digitizing Historical Pictorial Collections for the Intemet Weshington,DC: Council on Library and Information Resourors, February 1998) ‘A Kenney and O, Rieger, Using Kodak Photo CD Tech nology for Preservation and Access, May 1998, www libra yoornellod/prescrvation/kodal/oaveedtn, 150 16067-1, Electronic scanners for photographic. images —Spatial resolution measurements: Part 1, Scanners for reflective media, 203 F Frey, “Digital Imaging for Photographic Collections: Foundations for Technical Standards” RLO DigiNews, 1 (2), December 1997, wwa.tlg.ora/preserydizinewa/dii- nnewsS.htual##com, F Frey, “Digitization of Photographic Collections,” Proceedings, IS&T 5th Annual Conference, Boston, May 1997, pp. 597-599. E Frog, “Diaitize to Prevorve—Photographic Collections Facing the Next Millennium,” Proceedings, IS&T 50th Annual Conference, Boston, May 1997, pp. 713-715, 1997. F Froy, “Digitization of Photographic Collections,” Fry High Resolution and Quality hnaging Il, SPIE Proceed- ‘ngs, 3025, February 1997, pp. 49-52 F Frey and 8, Sisstrunk, “Image Quality sues for the Digitization of Photographic Collections,” Proceedings, IS&T 49" Annual Conference, Minneapolis, MN, May 1996, pp. 349-353. L, Stroohel and R. Zakia, eds., The Focal Ereyolopedta of Photography Boston; Focal Press, 1992), G.A. Geocheider, Pychophyaics—Method, Theory, and Applieation, 2 ed, London, Hilldale, NI: Lawrence Erlbaum Asociates, 1985), Y. Kipman, “Image Quality Metrics for Distal Image Capture Devices,” Proceedings, IS&T 48" Arousal Con- ferawce, Weshingion, DC, May 1995, pp. 456-459. P Engeldrum, ‘A Framework for Image Quality Mod- cls,” Journal of maging Science and Technology, 39 (4), 1995, pp. 312-819. P.Engeldrum, Introduction to Image Quality, short coarse notes, IS&T'11" International Congress on Ad- vances in Non- Impact Printing Technologies, November 1998. R. Shaw, ed, Selected Readings in Image Evaluation (Washington, DC: Society of Photographie Scientists and Engineers, 1976).S Herr, “Scanning for Gold,” Publish, December 1996, R. Gann andN. Shepard, Revieuing and Testing Desktop Scarmers, (Greeley, Colorado: Hewlett Packard ‘Company, 1997). RD. Forkert, etal, Zt Procadures for Ferifying [AFIS Scanner Image Quolity Requirements, MITRE Report for FBI Fingerprint Project, November 1994 A Kenney and 8. Chapman, Digital Resolution Rezuire- ‘ments for Replacing Text-Based Material: Methods for Benchmarking Image Quality (Washington, DC: Come mission on Preservation and Aoceas, 1995) ANSLAIMMSH-1988, Recommended Practice for Quality Control of Image Scanners (Silver Spring, MD: Association for Information and Image Management, 1988) M.Bster, Diztal Image Collections: Issues and Practice, (Wachingion, DC: Commission on Preservation and Ac- cose, 1996), M. Bator, “Specifies of Imaging Practice,” Archives & Maseton Informatics, 9, 1995, pp. 147-185, 180 12234-2, Photography-—Electronic Stil Pcture Cameras—Removable Memory—Part 2: TIRR/EP Image Data Format, 2001 ‘Tochnical Guidelines for Digitizing Archival Materials for Blectronic Access: Creation of Production Master Files—Raster Images, June 2004, www:archives gov/te- search/arc/digitizing-archival-material html 150 14524, Photography—Elostronio Still Piture Cameras—Methods for Measuring Opto-Electronie Consersion Functions (OBCFs), January 1989. ANSIAIM TR26-199, Resolution as it Relates to Pho- togrophic and Electronic Imasing (Silver Spring, MD: Association for Information and Image Management, 1993), P Barten, MTR OSB and SQRL for Image Quality Analysis, short course notes, ISWT/SPIE, Symposium ‘on Electronic Imaging: Science & Technology, January 1996. D.J. Braunegg, R.D. Forkert, and N. B. Nill Rescal- ing Digital Fingerprints: Techosiques and Image Quality Effects, MITRE Report for FEI Fingerprint Project, June 1996. Digital Imaging for Photographic Collections: Foundation » M.A. Chambliss, JA Dawson, and E. J. Bons, “Measue- ing the MTF of Undersampled IRFPA Sensors Using 2D Discrete Transforms,” SPIE Proceedings: Infrared Imag ing Systems, Design, Analysis, Modeling, and Testing VT, G.€ Hobs, ed, 2470, January 1995, pp. 312-324, 8. E. Reichenbach, etal, “Characterizing Digital Image Acquisition Deviers,” Optical Engineering, 30 2), 1991, pp. 120-17. HS. Wong, “Effect of Knife-Eelge Skew on Modulation ‘Transfer Function Measurements of Charge-Coupled Devise Imagers Employing a Scanning Knife Eelge,” Optical Engineering, 309), September 191, pp. 1394-1398, D. Williams, “Whatis an MTF .. and! Why Should You Care?” RLG DigiNews, 2 (I), February 1998, wows, ong) preserwdiginews/ dizinowe21 htmil##technical TA, Ficherand J. Holm, “Electronic Still Picture Camera Spatial Frequency Response Measurement,” Proceedings, IS&T 42" Annual Conference, Rochester, NY May 1994, pp, 626-630. R. Lamberts, “Ue of Sinusoidal Test Pattems for MTF. Evaluation,” technical information provided by Sine Paticrns LLC, Rochester, NY. 150 12233, “Photography—Electronic Still Picture Cameras—Resolution Measurements,” 200 D. Williams, “Benchmarking ofthe ISO 12283 Slanted- edge Spatial Frequency Response Plus-in, Proceedings, IS&T 1998 PICS Conference, Portland, OR, May 1998, pp. 133-137. A. Gelbart, Comparison of Sine Pattern cand Knife Ede Tchniques for Determining MTF of Distal Searners, B.S, Thesia, Rochester nattute of Technology, February 1997. N.B.Nill and B. R. Paine, Computer Program to Determine the Sine Wave Modulation Transfer Ranction of lmage Scanners, METRE Report for FBI Fiwzerprint Project, September 1994. 180 15739, Photography-—Elestronie Stil Picture Imag ing —Noise Measurements, 2003. J. Holm, “Log NEQ Baced Pictorial Print Noise Charac- terization,” Proceedings, IS&T 4°" Annual Conference, Rochester, NY, May 1994, pp. 429-432, J. Holm and S. Sisstrunk, “An BIQ-Subjective Image Standard:Quality Correlation Study” Proceedings IS&T 47* Annual Conference, Rochester, NY, May 1994, pp. 634-640, EF Frey and S, Sisstrank, “Color Issues to Consider in Pictorial Image Data Bases,” Proceedings, IS&T Fifth Color Imaging Conference, Scottsdale, AZ, November 1997, pp. 12-115 M.D. Pairchild, “Some Hidden Requirements for Device-Indlependent Calor Imaging,” paper presented at Society for Information International Symposium, 1994 R. Poe, “Aesthetic Considerations in Tone and Color Management,” Proceedings, IS&T Third Color Image ing Conference, Scottsdale, AZ, November 1995, pp 164-168, F Frey, R. Gschwind, and L. Rosenthaler, “Electronic Imaging, a Tool for the Reoonstruction of Faded Color Photographs and Motion Pictures,” Procerdings, SST" Fourth Cotor Imaging Conference, Noveraber 1996, pp. 59-44 F Frey and R. Gschwind, “Mathematical Bleach- ing Models for Photographic Three- Color Materia,” Joumal of imaging Science and Technology: 38 (5), November/December 1994, pp. 513-519. 180 17321/HD, Graphic Technology and Photogra- hy—Colour Characterisation of Digital Stilt Cameras (DSCs), Part 1, Stir, metrology and test procedures; Part 2, Methods for determining transforms from rate DSC to scene-refered data, 2008, 'S. Sisstrunk, “Imaging Production Systems at Corbis Comporation”, RLG DigtNews, 2 (4), August 1998, www. rigsong/presery/diginews/diginews2-4.htmif#technical, Data Dictionary —Technical Metadata for Digital Still Images, 2006, www nso. J. Holm, “Factors to Consider in Pictorial Digital Imase Prosossing,” Proceedings, IS&T 49" Annual Conference, Minneapolis, MN, May 1996, pp. 298-204. M. Rabbani, “Image Compression Fandamentals,” The Compression Experience: Proceedinas of The Rochester Section of the SMPTE Titorial, Rochester, NY, October 28, 1995, pp. 724. H, Kinoshita and'T. Yamamure, “Bifects of Repetitive PEG Compressions with DCT Block Rearrangement Operation on Image Quality” Jounal of Imaging Science and Technology, $9 (6), Noveniber/December 1995, pp. 546-558,sgraphic Collection fe BIBLIOGRAPHY G.A. Baxes, Digital Image Processing (New York: Jol Hiley & Sons, fac, 1994) R. Berns and F Frey, Direct Digital Capture of Cultural Heri- ‘tageBenchmarking American Museum Practices and Defin- ing Future Needs Direct Digitad Capture of Cultural Heritage — Benchmarking American Musetam Practices and Defining Radure Needs, wwv.cis rt.eda/museumSurvey/documents! Benchmark_Final_Report_Web.pdl, August 2005. C. Wayne Broun and B. J. Sheperd, Graphics File Formats, Refer- ence and Guide (Upper Saddle River, NJ: Prentice Hall, 1995), J.C Dainty and R. Shaw, Image Seience, Principles, Analysis, and Ezalaation of Photographic-Tspe Imaging Processes (London: Acaclemic Press, 1974) RG. Gann, Desktop Scanners (Upper Saddle River, NJ: Prentice Hall PTR, 1999). D. Hazen, J. Horrell, and J. Merrll-Oldham, Selecting Research Collections for Digitization Washinston, DC: Commaission on Preservation and Access, 1998). J. Holm and N. Judae, Electronic Photography at the NASA Langely Research Center” Proceedings, IS&T 48° Arewal Conference, Washington, DC, 1995, pp. 436-441 H. Maitre, FJ. M. Schmitt, and J. Crettes, “High Quality Imaging in Museums from Theory to Practice,” Fery High Resolution cand Quality Imaging If, Proc, SPIE, 3025, 1997, pp. 80-39, E. Murphy. Review of Standards Defining Testing Procedures for Characterizing the Color and Spatial Quality of Digital Camer- 96 Used to Image Cultural Heritage, wwe.ci.rit-edw/museum- Survey/documenta/StandardsReview_tp.PDF, 8. B, Ostrow, Disitising Pictorial Collections for the Internet, (Washington, DC: Commission on Preservation and Access, 1998). J. 6. Russ, The Image Processing Handbook, FL: ORG Press, 1995) A. Smith, Wy Digitize? (Washington, DO: Commission on Preser- vation and Access, 1999}, 24 od. (Boca Raton, C. Stevenson and P. McChung, ede, Delivering Digital hmages— Cultural Heritage Resources for Education. (Las Angeles, CA: Getty Information Institute, 1998)Fie SELECTED INTERNET RESOURCES Countless reeources and references can be found on the In- ternet. The following list ie a selection to start with. (World, Wide Web URLs confirmed ae of February 1, 2006.) Art Spectral Imaging artes.om Arts and Humanities Data Service ‘wrrw.ahds.ac.uk Council on Library and Information Resources www clinone Dublin Core ublincore.org Electronie Still Picture Imaxing vwwor iS. ongfit10.htral European Commission on Preservation and Asooss wwe knaw.nl/ecpa ‘The Intemational Color Consortium (CC) www.colororg, 180 Standards webstore.ansiong/ansidoestore/iso asp 180 Standards Tools ‘ww ida.ong/pdl/resource_order_form pdf 180 Tost Charts ‘wovwida.orglio_test_charta html Library of CongreseAmerican Memory—Technival Information, ‘memory oc gov/ammem/about/techn.himl MGN—Mazeum Computer Network swwvemon.edh ‘The National Archives www archives gov National Media Lab wwe nml.ong, LG DigiNews ‘wewrig.org/preserv/diginews ‘The Visual Resources Association www vranchong Selected Internet Resources 4THE IMAGE PERMANENCE INSTITUTE ‘The Image Permanence Inatitute is an academic research laboratory located on the campus of the Rochester Institute of Technology m Rochester, New York, Since its founding in 1985, IPI’s mission has been research for the advancement of the permanence and preservation of imaging media and information resources, IPLie cosponsored by RIT and the Society for Imaging, Science and Technology. Research at IPI deals primarily with preservation of images and recorded. information. IPI hae achieved suecess with projects involving enclosure qual- ity, silver image stability, decomposition of cellulosic plastic film supporte, color dye fading, paper deterioration due to air pollutants, magnetic tape preservation, and environmental assessment and control. IPI is known for its accelerated-aging studies of photographic materiale including acetate, nitrate, and polyester films, color dyes, and gelatin. These studies have underscored the strong role that environment plays in all modes of decay and the impor- tance af managing etorage for preservation. Among the products and zervices developed by IPI are the Photographic Activity Test, a worldwide standard (ISO Standard 18916) for archival quality in photographic enclosures; A-D Strips, a test for vinegar syndrome in acetate film, the Preservation Environment Monitor®, a unique electronic data logger designed specifically for preservation use; Climate Notehook® environmental analysis software; and the cost-effective Environmental Analysis Service for Aibranies, archives, and museums. IPI serves the prezervation community net only through its research, products, and services, but also as a ready source of technical information. In addition, IPI provides technical and administrative support for the American National Standards Institute (ANSI) and the International Organization for Standardization (ISO) in the area of permanence and care of imaging media, inclucling photographic materials, tape, and optical discs.