A Paper Presentation On DIGITAL IMAGE PROCESSING

Abstract: Over the past dozen years forensic and medical applications of technology first developed to record and transmit pictures from outer space have changed the way we see things here on earth, including Old English manuscripts. With their talents combined, an electronic camera designed for use with documents and a digital computer can now frequently enhance the legibility of formerly obscure or even invisible texts. he computer first converts the analogue image, in this case a videotape, to a digital image by dividing it into a microscopic grid and numbering each part by its relative brightness. !pecific image processing programs can then radically improve the contrast, for example by stretching the range of brightness throughout the grid from blac" to white, emphasizing edges, and suppressing random bac"ground noise that comes from the equipment rather than the document. #pplied to some of the most illegible passages in the Beowulf manuscript, this new technology indeed shows us some things we had not seen before and forces us to reconsider some established readings.

Introduction to Digital Image Processing $ision allows humans to perceive and understand the world surrounding us. %omputer vision aims to duplicate the effect of human vision by electronically perceiving and understanding an image. &iving computers the ability to see is not an easy tas" ' we live in a three dimensional ()*+ world, and when computers try to analyze ob,ects in )* space, available visual sensors (e.g., loss of information. .n order to simplify the tas" of computer vision understanding, two levels are usually distinguished/ lo!"le#el image processing and $ig$ le#el image understanding.

$ cameras+ usually give two dimensional (-*+

images, and this pro,ection to a lower number of dimensions incurs an enormous

0sually very little "nowledge about the content of images 1igh level processing is based on "nowledge, goals, and plans of how to achieve those goals. #rtificial intelligence (#.+ methods are used in many cases. 1igh' level computer vision tries to imitate human cognition and the ability to ma"e decisions according to the information contained in the image.

his course deals almost exclusively with low'level image processing, high level in which is a continuation of this course. #ge processing is discussed in the course Image Anal%sis and &nderstanding, which is a continuation of this course.

'istor% 2any of the techniques of digital image processing, or digital picture processing as it was often called, were developed in the 3456s at the 7et 8ropulsion 9aboratory, 2. , :ell 9abs, 0niversity of 2aryland, and few other places, with application to satellite imagery, wire photo standards conversion, medical imaging, videophone, character recognition, and photo enhancement. :ut the cost of processing was fairly high with the computing equipment of that era. .n the 34;6s, digital image processing proliferated, when cheaper computers %reating a film or electronic image of any picture or paper form. .t is accomplished by scanning or photographing an ob,ect and turning it into a matrix of dots (bitmap+, the meaning of which is un"nown to the computer, only to the human viewer. !canned images of text may be encoded into computer data (#!%.. or E:%*.%+ with page recognition software (O%<+.
(asic Concepts

# signal is a function depending on some variable with physical meaning. !ignals can be o One'dimensional (e.g., dependent on time+, o wo'dimensional (e.g., images dependent on two co'ordinates in a plane+, o hree'dimensional (e.g., describing an ob,ect in space+, o Or higher dimensional.

Pattern recognition is a field within the area of machine learning. #lternatively, it can be defined as =the act of ta"ing in raw data and ta"ing an action based on the category of the data= >3?. #s such, it is a collection of methods for supervised learning. 8attern recognition aims to classify data (patterns+ based on either a priori "nowledge or on statistical information extracted from the patterns. he patterns to be classified are usually groups of measurements or observations, defining points in an appropriate multidimensional space. #re to represent, for example, color images consisting of three component colors.

Image )unctions:

he image can be modeled by a continuous function of two or three variables/ #rguments are co'ordinates x, y in a plane, while if images change in time a third variable t might be added. he image function values correspond to the brightness at image points. he function value can express other physical quantities as well (temperature, pressure distribution, distance from the observer, etc.+. he brig$tness integrates different optical quantities ' using brightness as a basic quantity allows us to avoid the description of the very complicated process of image formation.

he image on the human eye retina or on a $ camera sensor is intrinsically -*. We shall call such a -* image bearing information about brightness points an intensit% image.

he real world, which surrounds us, is intrinsically )*. he -* intensity image is the result of a perspective pro,ection of the )* scene. When )* ob,ects are mapped into the camera plane by perspective pro,ection a lot of information disappears as such a transformation is not one'to'one.

<ecognizing or reconstructing ob,ects in a )* scene from one image is an ill' posed problem. <ecovering information lost by perspective pro,ection is only one, mainly geometric, problem of computer vision. he second problem is how to understand image brightness. he only information available in an intensity image is brightness of the appropriate pixel, which is dependent on a number of independent factors such as
o

Ob,ect surface reflectance properties (given by the surface material, microstructure and mar"ing+, .llumination properties, #nd ob,ect surface orientation with respect to a viewer and light source.

o o

Digital image properties

Distance is an important example. he distance between two pixels in a digital image is a significant quantitative measure. he Euclidean distance is defined by Eq. -.@-

Cit% bloc* distance

C$essboard distance Eq. -.@@

Pi+el ad,acenc% is another important concept in digital images. -"neig$bor$ood ."neig$bor$ood .t will become necessary to consider important sets consisting of several ad,acent pixels '' regions.

<egion is a contiguous set.

%ontiguity

paradoxes

of

the

square

grid

One possible solution to contiguity paradoxes is to treat ob,ects using @' neighborhood and bac"ground using A'neighborhood (or vice versa+. # hexagonal grid solves many problems of the square grids ... any point in the hexagonal raster has the same distance to all its six neighbors. (order < is the set of pixels within the region that have one or more neighbors outside < ... inner borders, outer borders exist. Edge is a local property of a pixel and its immediate neighborhood ''it is a vector given by a magnitude and direction.

he edge direction is perpendicular to the gradient direction which points in the direction of image function growth. (order and edge ... the border is a global concept related to a region, while edge expresses local properties of an image function.

Crac* edges ... four crac" edges are attached to each pixel, which are defined by its relation to its @'neighbors. he direction of the crac" edge is that of increasing brightness, and is a multiple of 46 degrees, while its magnitude is the absolute difference between the brightness of the relevant pair of pixels. (Big. -.4+

Topological properties o) digital images

opological properties of images are invariant to rubber s$eet trans)ormations. !tretching does not change contiguity of the ob,ect parts and does not change the number One such image property is the Euler''8oincare characteristic defined as the difference between the number of regions and the number of holes in them.

%onvex hull is used to describe topological properties of ob,ects. r of holes in regions. he convex hull is the smallest region which contains the ob,ect, such that any two points of the region can be connected by a straight line, all points of which belong to the region.

&seses # scalar function may be sufficient to describe a monochromatic image, while vector functions are to represent, for example, color images consisting of three component colors. CONCL&SION Burther, surveillance by humans is dependent on the quality of the human operator and lot off actors like operator fatigue negligence may lead to degradation of performance. These factors may can intelligent vision system a better option. As in systems that use gait signature for recognition in vehicle video sensors for driver assistance.