(iotial image Processing )——— Gn ee es Explain the applications of digital image Processing, (RTU.2017 ‘Ans. Applications of Digital Image Processing : Some of the major fields in which digital image proctssing is widely used are mentioned below: © Image sharpening and restoration + Medical field + Remote sensing + Transmission and encoding + Machine/Robot vision * Colorprocessing . Pattetn recognition pee 5 Soe Sharpening wd Restoration + Image sharpening and restoration refers here to process images that have been captured from the modern camera to make them s better image or to manipulate those images in way f0 achieve desired result. It refers to do what photoshop usually does, ‘This includes zooming, blurring, sharpening, gray scale to color conversion, detecting edge’ and vice versa, image retrieval and image recognition. x Medi Fl Ts cine pte XDI fo the field of femote sensing, the area of the earth is scanned by a satellite or from a very high ground and then itis analyzed to obtain information about it. One particular application of digital image processing Dicitat: Image: Paocessine : INTRODUCTION TO in the field of remote sensing is to deteot infrastructure damages caused by an-carthquake.. 2... fertile eer ee procedure, So a solution to this is found in processing. An image of the effected area is captured from the above ground.and then it is analyzed to detect the various types of damage done by the earthquake, ‘The key steps include inthe analysis. © — The extraction of edges * Analysis and enhancement of various types of edges “Teassmiasion and Encoding s The very first fago that ‘bas been transmitted over-the wite wasifapin, 10 New York via a submarine cable. was ‘sent is shown below. The picture that was sent took three hours to reach from one place to another. Now just iniagine, that today we are able to see live video feed, or live cctv footage from one continent to another with just a delay ofseconds. It means that alot of ‘work has been done in this field too. This field does not only focus on transmission, but also on encoding: Many different formats have been’ for high or low bandwidth to encode photos and then stream it oer the internet or etc. MachineMRobot Vision: Apc mb many calleges tht a robot face today, one ofthe Lae we rae rer things, identify them, identify the iirdles etc. Much work has been contributed by this field and a complete other field of computer vision has been introduced to work on et Scanned with CamScanner2 ‘Hardie Detection : Hurdle detection is one of the common. age processing, by calculating the distance betweersrebot and hurdles, Line Follower Robot : Most of the robots today work by following the fine and thus are called line follower robots. its path and perform a dea Vehieved through ae Cae Color processing includes processing cof colored images and different CSlor spaces that are used. Forexarnple RGB colormode!, YENCr, HSV.Tralso iavolves studying transmission, storaé,' of these.color imagen. 76 peter oe Pattern Recognition : ‘involves study Patterrreeotmition esa om (nie ine includes machine learning'(a branch of at intelligence). In pater recognition, image processing is used.for identifying the objects nan images and then * Video Processing : A video is nothing but just the very fastmorementof pictures. The quality ofthe video depends ‘on the tiumber of ftames/pictures per minute-and the quality of each frame being used: Video processing ° involves noise reduction, detail enhancement, motion detection, frame rate conversion, aspect ratio conversion, colorspace conversion ec, at 1s. Digital Image Processing? Give fatdameal steps in DIP. Explain each block. val a REU.2010) . Define the image. Explain the steps of gal Image processing with suitable diagram. ah ee - {RTU. 2017) OR. « Give fundamental steps in Digital Image Processing. Explain each block. [RI.V.2013] wader acca te ‘An image may defined 2s.atwo-dimensional of a digital computer. A digital image is composed of finite number of elements, each of which has. particular f- location and value, These elements are referred to as picture clement, image elements, pals and pels Pixelis the term most widely used to denote the elements of a igital image. sion is the most advanced of our senses, so it is not surprising that images play the single most important role in human perception. However, unlike humans, who are limited tothe visual band of the electromagnetic (EM) spectrum, imaging machines cover almost the entire EM spectrum, ranging from gamma to radio waves. They can operate on images geiérated by sources that humans are not accustomed to associating with images. These include ultra-sound electron microscopy and computer generated images. Thus digital image processing encompasses a ‘wide and varied field of applications. There areno clear-cut boundaries inthe continuum from image processing at one end'to computer vision at the other. However, one useful paradigm is to consider three types of computerized processes in this continuum; | low, mid and high-level processes. Low-level processes involve primitive operations such as image preprocessing to reduce noise, contrast enhancement and image sharpening, A low-level process is characterized by'the fact that both it inputs and outputs are images, Mid-level processing on images involves tasks such as segmentation (partitioning an image into regions or objects), description of those objects to reduce them to a form suitable for ‘computer précessing and classification (recognition) of individual objects. A mid-level process is characterized by the fact that its inputs generatly are images, but its outputs are attributes extracted from those images (e.g., edges, contours and thé identity of individual objects). Finally, higher-level processing involves “making sense” ofan ensemble of recognized objects, asin image analysis and at the far end of the continuum, performing the cognitive functions normally assoriated with vision, Fundamental Steps in Digital Image Processing () The acquisition could be as simple as being siven an image that's already in digital form. Generally, {he image acquiston sage involves preprocessing sich asscaling, (i) Image enhasicement is among the simplest and most appealing areas of digital image processing. Scanned with CamScannerthe idea behind enhancement techniques is to tring out detail that is obscured or simply to highlight certain features of interest in an image, A familiar example of enhancement is when we increase the contrast of an image because “it looks better.” pete nae tena retin tin oon L Fig. : Fundamental steps in digital Image processing (Gi) Image restoration is an area tat also deals With improving the appearance of an image, However, unlike enhancement, which is subjective, image restoration is objective, inthe Sense that restoration techniques tend to be based on mathematical or probabilistic models of image degradation. Enhancement, on the other hand is based on human subjective preferences regarding what constitutes a “good” enhancement result (iv) Color image processing is an area that has been gaining in importance because of the significant increase in the use of digital images over the Internet. ___(v) Wavelets are the foundation for representing images in Various degrees of resolution. (vi) Compression, 2s the name implies, deals with techniques for reducing the storage required to save an image or the bandwidth required to transmit it. Although storage technology hs improved significantly over the past decade the same cannot be sad for transmission capacity. This is tre particularly in uses ofthe Internet, which are characterized by significant pictorial content. Image ‘compression is familiar (perhaps inadvertently) to most ‘users of computers in the form of image file extensions, such as the jpg file extension used in the JPEG (Joint Photographic Experts Group) image compression (vil) Morphological processing deals with tools for extracting image components that are useful in the representation and description of shape. (vi Segmentation posedites prin damage 2) into its constituent parts or objects. In general, autonomous segmentation is one ofthe most difficult tasks in digital image processing, A rugged segmentation procedure brings. the process a long way toward successful solution of ifnaging problems that require objects to be identified individually. On the other hand weak or erratic, segmentation algorithms almost always guarantee eventual failure, In general, the more accurate the segmentation, the more likely recognition is to succeed (ix) Representation and description almost always follow the output of a segmentation stage, stage which usually is raw: pixel data, constituting either the boundary of region (i.e. the set of pixel separating one image region from atjother) or all the points inthe region itself. nether case, converting the data toa form suitable for computer processing is necessary. The first decision that must be made is whether the data should be represented as a bout i Boundary representation is appropriate when the focus is ‘on external shape characteristics, such as corners and inflections, Regional representation is appropriate when the focus is on-internal properties, such as texture or ‘skeletal shape. In some applications, these representations ‘complement each othr. Choosing a representation is only part ofthe solution for transforming raw data into form suitable for subsequent computer processing. A method must also be specified for describing the data so that features of interest are highlighted, Description, also called feature selection, deals with extracting attributes that result in some quantitative information of interest or are basic for differentiating one class of objects from another. (x) Recognition is the process that assigns a label (eg, vehicle”) to an object based on its descriptors. We conclude’ our coverage of digital image processing with the development of methods for recognition of individual objects. ‘ Knowledge about a problem domain is coded into ‘an image processing system in the form ofa knowledge database. This knowledge may be as simple as detailing regions of an image where the information of interest is ‘known to be located, thus limiting the search that has to be conducted in seeking that information, The knowledge base also can be quite complex, such as an inter-related list of all major possible defects ina materials inspection problem or an image database containing high-resolution ‘atellite-images of a region in connection with change- ‘detection applications. In addition o guiding the operation of each processing module, the knowledge base also controls the interaction between modules. This distinction is made in fig, by the use of double-headed arrows Scanned with CamScannerG@o——___ between the processing modules and the knowledge bas ul lowledge base, oppued {o single-headed arrows linking the ohne ___tisimportantto keep in mind that viewi of image pocesing ane pace ae ouput of Stage in fig. We also note that not all image processing applications require the complexity of interactions implied by fig. In fac, not even all those modules are needed i Some cases. For example, image enhancement for human visual interpretation seldom requires use of any of the other Stages in fig. In general, however, as the complexity of an image processing task increases, 50 does the number of processes required to solve the problem, peer ete ee eee ea G3 (0) Explain image sensing and acquisition. (0) Explain color vision modet with example. (RTU.2017| Ans.(a) Image Sensing and Acquisition : There are 3 pri ‘proportional to light intensity) (i) Single imaging sensor (Gi) Line sensor (ii) Array sensor ‘Image Acquisition using a Single Sensor : The most ‘common sensor ofthis type isthe photodiode, which is constructed of silicon materials and whose output voltage waveform is proportional to light. The use of a filter in front ofa sensor improves selectivity For example, a green (pass) filter in front ofa light sensor favours light in the ‘green band ofthe color spectrum. As a consequence, the sensor output wil be stronger for green light than for other inthe visiblespectrum. Tn order to generate a 2-D image using a single sensor, there have tobe relative displacements in both the x and y-directions between the sensor and the area to be imaged. An arangement used in high precision scanning, where a film negative is mounted onto a drum whose mechanical rotation provides displacement in one dimension, The single sensor is mounted on a lead screw that provides motion inthe perpendicular direction. Since ‘mechanical motion canbe controlled with high precision, this method is an inexpeasive (but slow) way fo obtain ‘ Ss flathed scanners. Sensing devices with 4000 or more in- Tine sensors are possible. line sensors are used routinely in airborne imaging applications, in which the finaging system is mounted onan aircraft that flies at» constant sensor arangements (produce an electrical output. tine wd speed ovethe geographical ae to beimaged. (ech, (Vi Sem) OS. Solved Papers (One-dimensional imaging sensor strips that respond to various bands.of the electromagnetic spectrum are mounted perpendigular to the direction of flight. The imaging strip gives one line ofan image at atime, and the motion ofthe strip completes the other dimension of a two-dimensional image. Sensor strips mounted in a ring configuration are used in medical and industrial imaging to obtain cross sectional (“slice”) images of 3-D objects. AA rotating X-ray source provides illumination and the portion of the sensors opposite the source collect the X- ray energy that pass through the object (the sensors obviously have to be sensitive to X-ray energy)-This is the basis for medical and industrial computerized axial tomography (CAT) imaging, Image Acquisition using Sensor Arrays : This type of, arrangement is found in digital cameras. typical sensor for these cameras is a CCD array, which can be manufactured with a broad range of sensing properties and can be packaged in rigged arrays of 4000 * 4000 ‘elements or more. CCD sensors are used widely in dig ‘cameras and other light sensing instrument. The response ‘of each sensor is proportional to the integral ofthe light energy projected onto the surface ofthe sensor, a property that is used in astronomical and other applications requiring low noise images. ‘The first function performed by the imaging system isto collet the incoming energy and focus tonto an image plane, Ifthe illumination i ight, the frontend ofthe imaging system isa lens, which projects the viewed scene onto the lens focal plane. The sensor array, which is coincident with the focal plane, produces outputs proportional to the integral of the light received at each sensor. ‘Ans.(b) Color Vision Model : Color vision model also known as color model or color space on color system) is the specification of colors in some standard and accepted way. Itis a specification of a coordinate system and & subspace within that system where each color is represented by a single point, The classification of color models can be done by (@) Hardware -oriented models. (i) Color description-oriented models: ‘The hardware-oriented models most commonly used are the RGB (Red, Green, Blue) model for color monitors and a broad class of color video cameras; the CMY (Cyan, Magenta, Yellow) and. CMYK (Cyan, ‘Magenta, Yellow and Black) models for color printing. models are HSI (Hue, Saturation and Intensity) model, ‘which describes and intéxpret different properties of colors. ‘The HSI model also helps decoupling the color and gray scale information of an image, Scanned with CamScannerRGB models thetnost widely used color model in “hardware application, in kis mode, each colorin an image Appear ts primar colo: component Red, Green, Blue The model follows cartesian coordinate system. Fig. : RGB Model The above shown figure is the subspace of RGB model. The primary values are positioned at the three axl comers ofthe cube, and the secondary color components are at the other three comers ofthe cube. Black is atthe origin and whit is at the farthest comer from black; and the line joining black and white comer is known as the sray-scale, The different color in this models are pointed ‘on or inside this cube, andthe value of particular points determined by the vector distance from the origin (Black) The general assumption is that all the color values are ‘normalized, so the cube shown in the above figure is a __nitoube, which signifies, all the color values will all inthe RGB is 24bit. This is known as full color image depth. Q4. What is image quantization? Explain the scalar ‘and image quantization in detail. (RIT.U.2016, {hy k= lyon L 1) a8 set of increasing transition or values, respectively, of u.IFulies in interval (t,t) ten itis mapped tor, the k* reconstruction level eet Fig. 1A quantier: Zero memory quantizers ae i techniques such es pulse céde modulation5{PCM), differential PCM, transform coding; and30 on Notethat the quantizer mapping is irreversible; that is, fora given quantizer output, the input value cannot be determined ‘uniquely. Hence, a quantizer introduces distortion, which any reasonable design method must attempt to minimize. ‘There are several quantizer designs available that offer various trade-offs between simplicity: 2 Scalar Quantization: Insealarqunizatio, the quantization oupatsthe result of division of the input data by a quantization parameter, ‘with rounding: tonite nero If X isan input sample and Q is a q the quantized output is xerten(3) There are different types of scalar lat cotton techniques uniform quantization, non-uniform quantization, and adaptive quantization. Uniform Quantization : Consider X.,, is the maximum value fom an input soured te inpt als sare uniformly distributed nie he an N-Jeyel uniform scalar quantizer Se tard XX] ae ee et Se ail enbrinra ma step sizeof the uniform quantizer ie Characteristics ofan 8-eve uniform sealar quantizer, is shown in Fig. 2. Horizontal axis represents the input Scanned with CamScannerandithe vertical axis represents the corresponding value aftetaquantization and inverse quantization, Hence any , value-in between (24, 34), say as an example, will be! approximated to 2.54. Uniform scalar quantization is very simple and straightforward for implementation. It is designed based ‘on assumption tha he input source is uniformly distributed But often probability of distribution of the source: symbols isnot uniform in nature andthe uniform scalar quantization results in poor reconstructed quality. As a result, there is ‘anecessity to design non-uniform quantizers for these types of sources, fees 40-30-00 -: ee) BN 3y 4y Input +250 =—4 -458 Figc2: Uniformsealar quantizer with step size A Non-uniform Quantization : In order to minimize the averagedistortion inthe reconstructed image because ofthe quantization, wecan lightly quantize the transform oefficients or prediction eror values in the region of high importance and heavily quantize the corresponding ‘coefficients in a less important region in the image. ‘One that the quantization steps are smaller for the samples ‘those have more concentration in the curve of probability distribution of the samples. For.example, dis tion of the prediction error values is more concentrated at the origin of the curve compared to prediction error values in, As.a result, we can use smaller teem {tothe other prediction error values, x, + X_] in the quantizer'is non- uniformly divided so that the relation between input and _ output fromthiéquantizr ean match to any desired linear Tac sell onion quate, Te charactetistic of a general non-uniform quantizer is shown inFi3sanexampe whee qanzaton ie pois (. y %) and the output levels (Y, : iin ere fixed and they can Y, minimize some fiction of the quantization eror ‘hen the probability distribution of the input is know Adaptive quantization. When the statistics of the Source symbols considerably changes in the process, the fixed and predefined uniform or non-uniform quantizer, fails to yield good results. tn this situation, the quantizers need to be adaptive with the changes of the source statistics. There are two classes of adaptive quantizers forward adaptive quantizers and backuvard adaptive izers (6). In a forward adaptive quantizer, the izet extracts the quantization step size from the input, In this approach the source data is divided into blocks, Each block is independently analyzed and the quantization step size is determined based on its statistical distribution, Ina backward adaptive quantizer, the quantization step size is determined based on the previously reconstructed ‘output signals from the quantizer. Both the forward and backward adaptive quantizers have their own advantages and disadvantages, Fig. 3: Charncterit of nomaniorm sai quantioer ee QS. Explain the concept of image representation and differentiate the image compression and representation. IRT.U2016} "i OR is enon Jeet oteeetes ot So IRTU.2011) ‘Ans, Image Representation and Modeling : In image representation one is concerned with characterization ofthe quantity that each picture-lement (also called pixel or pel) represents. An image could represent luminances of objects in a scene (such as . pictures taken by ordinary camera), the absorption haracteristics of the body tissue (X-ray imaging), the radar cross section of a target (radar imaging), the temperature profile ofa region (infrared imaging), or the perianal field in an area (in geophysical imaging). ln general, any two dimensional function that bears Scanned with CamScannerfortiation can be contidered an image, Image models ive logical or qui ive description ofthe properties * ofthis function, Figure | lists several image representation ‘id modeling problens: important consideration in image representation is the fidelity or intelligibility criteria for measuring the {quitity of an image or the performance of a processing tecttnique. Specification of such measures requires models ‘of perception of contrast, spatial frequencies, eolor and d oh. Knowledge of a fidelity criterion helps in designing, the’image sensor, because it tells us the variables that Should be measured most accurately. The fundamental requirement of digital processing is that imiages be sampled and quantized, The sampling. rate (rlumber of pixels per unit area) has to be large enough to preserve the useful information in an image. It is determined by the bandwidth of the image. For example, the bandwidth of raster scanned common television signal isabout 4MHz, From the sampling theorem, this requires a minimum sampling rate of MHz. At 30 frames/s, this ‘means each frame should contain approximately 266,000 pixels. Thus, fora $12 line raster, this means each image frame contains approximately $12 x $12 pixels. Image quantization isthe analog to digital conversion of a sampled image to finite number of gray levels. lew norte emg gure, 6 eo na eo eo ie Seen ee Siete ae “Sate ears a _ oe =— Fig. | : Image representation and modeling, ‘A classical method of signal representation is by ‘an orthogonal seies expansion, such asthe Fourier series. Forimages analogous representation is possible via two- dimensional othogonal functions called basis images. For cee ups tts iageciabe ceed et ‘unitary matrices called image transforms. Any given image ~ ean be expressed as a weighted sum ofthe bass images equation and forced by white noise or some other random field with known power spectrum density is a useful approach for representing the ensemble. Fie. 2: bmeserpeentaon orthogonel basis image series Fig.3 shows three cs €5 of stochastic models where ‘an image pixel is characterized in terms of its neighboring pixels. Ifthe image were scanned top to bottom and then leftto right, the model of Fig. (a) would be called causal model. This is because the pixel A is characterized by pixels that lie in the “past.” Extending thi idea, the model ‘of Fig.3 (b) isa non-causal model because the neighbors of A lie in the past as well as the “future” in both the directions. In Fig.3 (¢), we have a semi-causal model because the neighbors ofA are inthe past inthe j-direction and are in the past as well as future in the i-direction, ripen fet Che isieeeh eats! nine be iNet fe io Aes io A oi A ite 4 Wee capil (0) Casal mode. (0) Wenenual model, (c)Semicnsl mode Fig. 3 : Three canonical forms of stochastic models Such models are useful in developing algorithms that have different hardware realizations. For example, causal models can realize recursive filters, which require small memory while yilding.an Infinite Impulse Response (UR). On the other hand, non-causal models can be used to design fast transform-based Finite Impulse Response (FIR) filters. Semi-causal models can yield two- dimensional algorithms, which are recursive in one -dimension end-non-recursive inthe other. Some of these stochastic models can be thought of as generalizations of ‘one dimensional random processes represented by Auto-Regressive (AR) and Auto-Regressive Moving ‘Average (ARMA) models. Difference between Image Compression and _ Representation: Image Compression : Compression, as the name implies, deals with techniques for reducing the storage required to save an image, or the bandwidth required to Scanned with CamScanner ee ae ee ae a a ae ee aeHanan Although storage technology has improved significantly oy r tne dant be ignficantly over the past decade, the same cannot be (rized by significant ssion is familiar (perhaps computers in the form of he jpg file extension used ie Experts Group) image Pictorial content. Image compre inadvertently) to most users of image file extensions, such a th in the JPEG (Joint Photograph Compression standard, : fe oe + Representation and description : ays follow the output ofa segmentation stage, Which usually is raw pixel data, constituting either the boundary ofa region (i. the set of pixels separating one image region from another) or all he points in the region itself. In either case, convert for computer processing is necessary. The first dec that must be made is whether the data should be Tepresented asa boundary or as a complete region. Boundary representation is appropriate when the focus is fon external shape characteristics, such as corners and inflections, Regional representation is appropriate when the focus is on internal properties, such as texture or skeletal shape. In some applications, these representations complement each other. Choosing a representation is only partof the solution for transforming raw data into a form suitable for subsequent computer processing, A method must also be specified for describing the data so that features of interest are highlighted. Description, also called feature selection, deals with extracting attributes that result in some quantitative information of interest or are basic for differentiating one class of objects from another. ————— Q6 Write short note on Representing digital images. IRT.U. 2012] OR What do you mean by Sampling and Quantization in Digital Image Processing. How digital Images are represented? |RLT.V.2013] Let f(s, 1) represent a continuous image function of two continuous variables, and1. We convert this function into a digital image by sampling and quantization, Suppose that we sample the continuous image into a 2-D array, lx, y), containing M rows and N columns, where (x, y) are discrete co-ordinates. For notational clarity and convenience, we use integer values for these discrete co-ordinates: x=0, 1, 2 M—1 and y=0, 1, 2yoyN= I, ‘Thus, for example; the value of the digital image at the ‘origin is (0, 0),nd the next co-ordinate value along the first row is (0, 1), Here, the noltion (0,1) is used to signify (a-Tech, (Vill Bem.) CS. Solued Papece the second sample along the first row. It does not médn that these are the values of the physical ¢o-ordinates when the image was sampled. In general, the value ofthe image at any co-ordinates (x, y) is denoted f(x,y), where x and Yate intergers. The section of the real plane spanned by the co-ordinates of an image is called the spatial domain, with x and y being referred to as spatial variables or spatial co-ordinates As Fig. shows, there are three basic ways to tepresent f(x,y), Figure 1(a) isa plot of the function, with two axes determining spatial location and the third axis being the values of intensities) asa function of the two spatial variables x and y. Although we can infer the structure of the image in this example by looking at the plot, complex images generally are too detailed and difficult to interpret from such plots. This representation is useful when working with gray-scale sets whose elements are expressed as triplets of the form (x, y, 2), where x and y are spatial co-ordinates and z is the value of f at co-ordinates (x,y). (b), © Fig. 1; (a) Image ploted as a surface, (b) Image displayed os @ visual intensity array, () Image shown ax a2-D numericleray (0,05 and Iepresen blac, gray and white, respectively) The representation in Fig.b (b) is much more common. It shows f(x, y) as it would appear on a monitor ‘or photograph. Here, the intensity of each point is proportional to the value of fat that point. In this figure, there are only three equally spaced intensity values. If the intensity is normalized tothe interval (0,1), then each point inthe image has the value 0, 0.5 or 1. A monitor or printer simply converts these three values to black, gray, ‘or white, respectively, as. Fig.1 (b) shows. The third representation is simply to display the numerical values of (x,y) 48 an array (matrix) In this example, fis of size Scanned with CamScanner60 +600 elements, or 360,000 numbers. Clearly, printing the:complete array would be cumbersome and convey litleinformation. When developing algorithms, however, ‘tis representation is quite useful when only parts of the image-are printed and analyzed as numerical-values. Fig. (¢) conveys this concept graphically. ‘We conclude that the Tepresentations in Fig.1(b) and 1(¢) are the most useful. Image displays allow us to View results at a glance Numerical arrays are used for processing and algorithm development. In equation form, we write the representation of an M x N numerical array as F(0,0)— f(0,1) F(0,N-1) (1,0) it N= flay)e| Joo). Huy ue ) sM- 1,0) foe 1) a f(M-LN=I), Al) Both sides of this equation are equivalent ways of expressing a digital image quantitatively. Te right side is ‘matrix of real numbers, Each element ofthis matrix is called an image element, picture element, pixel, or pel. The terms image and pixel are used throughout the book to denote a digital image and ts elements. In some discussions it is advantageous to use a ‘more traditional matrix notation to denote a digital image and its elements: Mo - %) Hy a ao. a N-1 : wn(2) ae AO MAM om IM AN ot. = x= i, y= j)= i,j), 80 Eqs. (1) and @) are Salas matrices. We can even represent an image as a vector, v. For example, a column vector of size MN x 1 is formed by letting the first M elements of v be the first column of A, the next M elements be the second column, and so on, Alternatively, we can use the ‘rows instead of the columns of A to: form’such a vector. Either representation is valid, as longas we are consistent. into digital form. This involves two processes: sampling and quantization. Digitizing the co-ordinate values + Digitizing the amplitude values is Basic Concepts in Sampling and Quantization The basic idea behind sampling and quantization is illustrated in fig.2(a) shows a continuous image, f(x,y), that we want to convert to digital form. An image may be continuous with respect to the x-and y-co-ordinates and also in amplitude. To convert it to digital form, we haveto sample the functions in both co-ordinates and in agpplitude. A 20) Fig. 2 : Generating a digital image (a) Continuous image, (b) A sean line from A to B in the continuous image, used to illustrate the concepts of sampling and quantization, (c) Sampling and quantization, (@) Digital scan line = SBE The one-dimensional function shown in fig. 2(b) is 4 plot of amplitude (gray level) values of the continuous image along the lin segrnent AB in fig 24a). ‘Variations are due to image noise. Tosa i ion, we take equally spaced samples al $3 in fig. 2(c). The location of each sample is given by 2 vertical tick mark in,the bottom part of the figure. The samples are shown as small white squares superimposed ‘on the function. The set of these discrete locations gives the sampled function, However, the values of the samples still span (vertically) a continuous range of gray-level values. In order to form a digital function, the gray-level values also must be converted (q into discrete ‘quantities. The right side of fig, 2(c) shows the gray-level.” scale divided Lika pretender to white, The vertical tick marks i assigned to each ofthepigit ew eae eens ‘gray levels are quantized simply by assigning’one of the ~ eight discrete gray levels to each campleto vertical tick * Scanned with CamScanner‘mark: The digital samples iesulting from both sampling, ‘and quantization are shown in fig. 2(d). Starting at the top Of the image and carrying out this procedure line by line produces a two-dimensional digital image. Saraplng inthe manner just described assumes that ‘we have a continuous image in both co-ordinate directions as wellasin ampliuide. In practice, the method of sampling. is determined by the sensor arrangement used to generate the image. When an image is generated by a single sensing element combined with mechanical motion. However, sampling is accomplished by selecting the number of individual mechanical increments at which we activate the sensor to collect data, Mechanical motion can be made ‘very exact so, in principle, there is almost no limits are established by imperfections in the optics used to focus on the sensor an illumination spot that is inconsistent with the fine resolution achievable with mechanical displacements, When a sensing strip is used for image acquisition, the number of sensors in the strip establishes the sampling limitations in one image direction, Mechanical motion in the other direction can be controlled more accurately, but it makes little sense to try to achieve sampling density in cone direction thatexceeds thesampling limits established by themimberof.sensors in the other. Quantization of the sensonoatputcomplétes the process of generating a digital image: Pape) Coeienr mage projected ont | Sensor array, (0) Resa of maze sampling and quantization : When sensing ary isused for image acquisition, there iso motion and the number of sensors in the array establishes the limits of sampling in both directions. ‘Quantization of the sehsor outputs is as before. Figure 3 illustrates this coacept. Fig (8) shows a continuous image ‘Projected onto the plane of an array sensor. Fig. 3(b) shows the image after sampling and quantization. Clearly, the pore mcr rameneen sect by: crete gray levels used in (B.Tech. (Vill Sem) CS. Solved Papers Q7 What do you mean by Aliasing in Digital Image Processing? Explain Moiré pattern, |RTU, 2013 201 Ans, Aliasing and Moire Patterns Functions whose area under the curve is finite gan be represented in terms of sine and cosines of various frequencies. The sine/cosine component with the highest frequency determines the highest “frequency content” of the function, Suppose that this highest frequency is finite and that the function is of unlimited duration (these functions are called band-limited functions). Theiv the Shannon sampling theorem tells us that, if the function is ‘sampled at a rate equal to or greater than twice its highest frequenoy, it is possible to recover completely the original funiction from its samples. Ifthe function is under sampled, then a phenomenon called aliased frequencies. Note that the sampling rate in images is the number of samples taken (inboth spatial directions) per unit distance, As itturns out, except for a special case discussed in the following paragraph, itis impossible to satisfy the sampling theorem in practice. We can only work with ‘sampled data that are finite in duration, We can model the process of converting a function of unlimited duration into a function of finite duration simply by multiplying the unlimited function bya “gating function” thatis valued | for some interval and 0 elsewhere. Unfortunately, this function itself has frequency components that extend to Hi ee tt i te ~ i ni _ i il emanate Figs Mustration ofthe Moire Potern Effect Thus, the very act of limiting the duration ofa band- limited function causes it to cease being band limited, which causes it to violate the.key condition of the sampling theorem. The principle approach for reducing the aliasing effects on an image is to reduce its high-frequency components by blurring the image prior to sampling. Scanned with CamScanner* However, fowever, aliasing is always present in a sampled image, Thie'effect of alinsed frequencies can be seen under the right conditions inthe forth of so called Moire patterns, "There is one special case of significant importance in'which a function of infinite duration can be sainpled ‘overa finite interval without violating the sampling theorem, ‘When a function is periodic, it may be sampled at a rate equal to or exceeding twice its highest frequency and itis possible to recover the function from its samples provided that the sampling captures exactly an iiteger number of periods of the function "This special case allows us to illustrate vividly the Moire effect. Fig. shows two identical periodic patterns of equally spaced vertical bars, rotated in opposite directions and then superimposed on each other ‘by multiplying the two images. A Moire pattern, caused by a breakup of the periodicity, is seen in fig. as a 2-D sinusoidal (aliased) waveform (which looks like a corrugated tin roof) running ina vertical direction. A similar pattern can appear when images are digitized (e.g. scanned) from a printed page, which consists of periodic ink dots. : Q8 Define sampling theorem and write a short note on Image quantization, {R-7.U. 2011] ‘Ans. Sampling Theorem A band limited image f{x, y) sampled uniformly on a rectangular grid with spacing Ax, Ay can be recovered ‘without error from the sample values f{mAx, ny) provided the sampling rate is yreater than the Nyquist rate, that is a y 7S Moreover, the reconstructed image is given by the interpolation formula f(x,y) bE f'(mAx, nby) Scanned with CamScanner a a