DIGITAL IMAGE PROCESSING MORPHOLOGICAL IMAGE PROCESSING Multiple Choice Type Questions 4. Erosion ; ‘JWBUT 2010) a) sharps a region . b) blurs a region c) increases a region ‘ d) decreases a region Answer: (C) ... is not a morphological image processing algorithm. [WBUT 2011]. a) Thinning b) Skeleton ¢) Both (a) & (b) d) None.of these Answer: (4) 3. Knowledge .of which one of the following is not required for morphological image processing? [WBUT 2011, 2012] a) Erosion b) Morphological reconstruction c) Neural networking 4) Duality & dilation Answer: (d) 4, Which of the following operations is idempotent? [WBUT 2016] a) dilation b) erosion ¢) convolution d) closing Answer: (d) 5, The morphological operation used to isolate objects which may be just touching one another is [WBUT 2017) a) dilation b) erosion c) opening d) closing Answer: (b) Short Answer Type Ques’ 1. What is boundary extraction in Morphological image operation? [WBUT 2014, 2017, 2018] Answer: When dealing with binary images, the principle application of morphology i image components that are useful in the representation and description of shape. The boundary (A) of a set A, f(A) = A-(40B). where B is a suitable structuring element. is extracting,POPULAR PUBLICATIONS Illustration WY A encoded by B Therefore, difference between the operation of an order statistics filter and classify it. [WBUT 2015) An 3 Order statistics filters are spatial filters whose response is based on ordering the pixel contained in the image area encompassed by the filter. The response of the filter at any point is determined by the ranking result. ; Median filter is the best order statistics filter; median of gray levels in the neighbourhood of the pixel. ie., f(x, y)=median (g(s.1)) it replaces the value of a pixel by the nIP.o4The original of the pixel is included in the computation of the median of the filter are quite possible because for certain types of random noise, thus provide excellent noise reduction capabilities with considerably less blurring then smoothing filters of similar size. These are effective for bi-polar and unipolar impulse noise. * ' a) Skeleton are the minimal representations of objects in an image w' taining the enter number of the image. The enter number is the number of objec! image minus the number of holes in those objects. The skeleton on of object ian age can be found by successive thining until stability is reached. ’b) Effect of dilation Process: 1. Dilation expands the input image. 2. This process effects both the inside and outside of the input image. 3. Dilation fills the holes in the image. 4. The dilation Process smoothens out the im ur. 3. a) Briefly explain skeletons. 16) b) Write down the effects of the dilation and erosion process. Answer: . 5: re Effect of erosion process: 1. The erosion process shrinks.th it image. 2. Pixels and holes inside the ima; jormally removed by the erosion process. 3. Erosion Process removes small objects like noise. . _4. Explain the lines jetecting process with mathematical models. [WBUT 2016] Answer: . Line masks are h , 45° line, vertical line, 45° line and is given by -l -l -1.2'-1 horizontal 2 2]; vertical line}-1 2 -1 Se | -1 2 -l -1 2 2-1-1 45° line}-1 2 -1]; -45°line}-1 2 -1 “ [2 -1 -1 -1--1 2 If at a certain point in the image, |R|>|R,| for all ji, that point is said to be more likely associated with a line in the direction of mask i, where R, & R, are the response of the mask. Edge detection locates sharp changes in the intensity function. Edges are pixels where brightness changes abruptly. A change of the image function can be described by a gradient that points in the direction of the largest growth of the image function. An edgePOPULAR PUBLICATIONS is a property attached to an individual pixel and is calculated from the image function behavior in a neighbourhood of the pixel. Magnitude of the first derivative detects the presence of the edge. Sign of the second derivative determines whether the edge pixel lies on the dark sign or lightside. Profile of a horizontal line First derivative Second derivative ve Fig: (a) Fig: (b) Fig. Edge detection by derivative operators. a) light stripe on a dark background b) dark stripe on a light background 1D, Type Questions 4. Explain the chain [WBUT 2010} Answer: 7 Chain codes are used to represent a boundary for connected sequence of straight line segments of sp gth and direction. Typically this representation is based on 4- or 8- connectivity of the segments. The direction of each segment is coded by using a mm shown below. 1 2 0 r , 3 A boundary code formed as a sequence of such directional members is referred to a Freeman Chain Code. Digital images usually are required and processed in a gris format with equal spacing in the x- and y-directions, so a chain code can be generate by. following a boundary in, say a clockwise direction and assigning a direction to th DIP. =DIGITAL IMAGE PROCESSING segments connecting every pair of pixels. This method generally is unacceptable for two principal reasons: - 1) The resulting chain tends to be quite long and 2) Any small disturbances along the boundary due to noise. or imperfect segmentation cause changes in the code that may not be related to the principal shape features of the boundary. An 2pproach frequently used to circumvent these problems is to resi le undary by selecting a larger grid spacing as shown-below: Then, as the bound: e assigned to each node of the large grid, depending on imity of the original boundary to that node, as shown in below. esPOPULAR PUBLICATIONS a 4- or 8-code. Fig. = tte tite ited by The resampled boundary obtained in this way can be represent al chain code. It is below shows the coarser boundary points represented by an 8-direction a simple matter to convert from an 8-code to 4-code and vice-versa. rN, ) i 1 ~~ \ me 2 = 2 g s ® s 3 € 5 a = 2 = = 6 S The starting point in the figure is at tl gives the chain code 0766 .... 12. As mi code representation depends on the s| expected, the accuracy of the resulting the sampling grid. . on the starting point. However, the code can be The chain code of a boundary depen : normalized with respect to the st: int by a straight forward procedure. We simply treat the chain code as a circt ence of direction numbers and redefine the starting point so that the resulting seques ‘of numbers forms an integer of minimum magnitude. We can normalize al tion by using the first difference of the chain code instead of the code i hi ference is obtained by counting the number of direction changes that sé adjacent elements of the code. For instance, the first difference of the 4-direction chain code 10103322 is 3133033. If we treat the code as a circular sequence cn i lize with respect to the starting point, then the first element of the difference mputed by using the transition between the last and first components of how that the first difference of a chain code normalizes it to rotation. b) Compute the first difference of the code 0110233210332322111, c) What is the order of the shape number for the figure shown? d) Obtain the shape number. oo ka DIP-98 the chaifiy Here the result is 33133030. Size normalization can be achieved by altering the si sampling grid. . s [WBUT 2012]DIGITAL IMAGE PROCESSING Answer: . 7 a) We can normalize for rotation by using the first difference of the chain code instead of the code itself. The difference is obtained by counting the number of direction changes . (in counter clockwise direction) that separate two adjacent elements of the code. For example, the first difference of the 4-diréction chain code 10103322 is 3133030. If we elect to treat the code as a circular sequence, then the first element of the di computed by using the transition between the last and the first compont The result is 33133030, . 1 by 0110233210332322111 5 | ‘ ‘The first difference will be \ 103210333303 130300 : by using above diagram. °) The order of the shape n‘ 4) Chain code 0030: Difference 0. 303130313 The shape no. will be smallest magnitude ie., _0313031303130313 "3, a) Wha mean by image recognition? IWBUT 2017] b) Des: iny image recognition technique. . . Ans a) Image recognition also known as pattern recognition, in the context of machin: is the ability of software to identify objects, places, people writing and action in images. Computers can use machine vision technologies in combination with a camera ‘and artificial intelligence software to achieve image recognition. 7 b) Image recognition is the process of identifying and detecting an object or a feature in digital image. This concept is used in many applications.- Typical image recognition algorithms include:POPULAR PUBLICATIONS Optical Character recogni Pattern matching Face recognition License plate matching Scene change detection Sm Pattern Matching: . ; at A Pattern is an arrangement of descriptors: The name feature is used o| aaa that recognition literature to denote a descriptor. A pattern, class is a where w is share some common properties. Patterns classes are denoted Wy, W ues for assigning the number of classes. Pattern recognition’ by-machine invo! us intervention as patterns to their respective classes automatically and witl le human i possible. Three common patterns are vectors, strings and trees. eecee 4. Write short notes on the following: ‘ T 2009) 2) Morphology : waur 2014, 20171 b) Hit-or-miss Transform [WBUT 2015, 2018] 3 Cannan Dilation [WBUT 2016, 2018) Answer: a) Morphology: ; . Morphology is a tool for extracti omponents useful in the representation and description of region. shape boundaries, skeletons and convex hulls. The language of mathematical mo is set theory and as such it can apply directly to binary images. Basic operations of mathematical morphology operate on two sets. The first one is the imag he second one is the structuring element sometimes called” Kemel.’The struct nt used in practise is generally much smaller than: the - image, often a 3 sion and dilation are two basic operators in mathematical morphology. b) Hit-or-miss Transform: The hi transform is a general binary morphological operation that can be used to lool ticular pattern of foreground and background pixels in an image. It is basic operation of binary morphology since almost all the other binary rat logical operators can be derived from it. As with other binary morphological lors it takes as input a binary image and a structuring element and produces another binary image as output. The structuring element used in the hit-and-miss is a slight extension to the type that has been introduced for erosion and dilation, in that it contains both foreground and background pixels rather than just foreground pixels i.e. both ones and zeros. The hit-and-miss operation is performed in much the same way as other morphological operators, by translating the origin of the structuring element to all points \ DIP-100DIGITAL IMAGE PROCESSING ., the image and then comparin; . the structuring element with the underlying image as | grosion and Dilation: ation and Erosion are two fundamet poundaries of objects i ‘ntal morphological i ‘lati i wt jects in an image, while erosien removes, piel we bjet oa jes. The number i: goundaries: | OF of pixels added : the si ed to rem jects i ages o" ize and shape of the s turing seen ee ie : " eH chain code: Refer to Question No. 1 of Long Answer Ty, er Type xe x S$ er * s @ OC) pilPOPULAR PUBLICATIONS IMAGE SEGMENTATION, REPRESENTATION AND DESCRIPTION Multiple Choice Type Questions eH 1. Representation & description almost always follow the outpu IT 2011, 2012, 2013) a) Segmentation stage b) Filtering stag . _ ¢) Compression stage 4d) All of thes Answer: (a) 2. Isolated point can be best detected using [WBUT 2014] a) 1*' derivative 2 ivative c) 3° derivative . e of these Answer: (b) 3. What operation is needed for b: removal? [WBUT 2015] a) Image multiplication b) Image addition c) Image subtraction 4d) all of these Answer: (d) 4. Representation a) segmentatio c) compre: Answer: (a) scription almost always follow the output of a 3 b) filtering stage [WBUT 2017} st 4d) all of these Short Answer Type Questions 7 exci the region growing technique for image segmentation and mention the associate with it. on (WBUT 2012, 2017] fly explain about region growing technique. (WBUT 2014) Answer: Region growing segmentation is an approach to examine the neighbouring pixels of the initial “seed points” and determine if the pixels are added to the seed point or not. Step 1: Selecting a set of one or more starting point (seed) often can be based on the nature of the problem. DIP-102DIGITAL IMAGE PROCESSING 2; The region are grown from: these seed points to adjacent point depending on a ste? ‘old or criteria (8-connected) we make. aoe gegion growth should stop when no more pixels satisfy the criteria for inclusion » tat reBiO"- Here suitable selection of seed point is important and more information of image is better. The result of an image after region growing still have points gray- ae higher than the threshold but not connected with the object in image. &S x write down the usefulness of segmentation? {WBUT 201 answer? is typi , sos etc.) ge segmentation is typically used to locate objects and boundaries (linés,, curves, etc. «images: More precisely image segmentation is the process of assignii 0 every jyel in an image such that pixel with the same label share certai teristic. Image _ entation is the process of dividing an image into multiple parts, This is typically wed to identify objects or other relevant information in digital images. 1 2 wr . 0 1 . (on 2 it v={0,1}. Compute the (i) shortest cency distance and (ii) m-adjacency distance between the pixels p and Answer? Wien v = {0,1}, 4-path does fom p to q by travelling alon «-Fig, (a) show this condi gah (shown dashed) i possibility for The shortest 4-path whi {1, 2} is shown in fig, (b): Its length is 6. It is easily verified tha another vO) the same length exits between p and q. 2 Iq -3 1 2. 1@) 2 2 2 0 20 consider the image segment 3 2 [WBUT 2017] onn= ‘it between p and q because it is impossible to get t are both 4- adjacent and also have valves from ot possible to get to q. The length of the shortest m- Both of these shortest paths are unique ix this case. One 1 2 0 t dow 14 1 1 2° 1 1 1 1 (Pl OT 2 (pla 0 (a) (b) DIP-103POPULAR PUBLICATIONS Long Answer Type Questions 1. a) What is. boundary descriptor? : b) Explain Fourier descriptor. : BUT 2010} Answer: a) There are several simple geometric measures that can be useful for boundary. The length of a boundary which is the number of pi: gives a rough approximation of its length. Curvature is the rate of change of slope. To measure a curyature accu! tely at a point ina digital boundary is difficult. The difference cen of adjacent boundary point of i Segments is used as a descriptors of curvature at the intersections of segments Order 4 Chain code 0321 ol First difference . 3331 03303 Shape no. 3333 33 .The shape number of a ry is defined as the first difference of a smalle magnitude. The order n ofa sl ber is defined as the number of digits in its representation. ‘ulated by treating the chain code as a circular sequence, Lt Chain code: 00332211 03032211 00032221 Difference: - 30303030 33133030 30033003 Shape no.: 03030303 03033133 00330033fourier descriptors are a way of encoding the shape of a two-dimensional object by raking the Fourier transform of the boundary, where every (x, y) point on the boundary is mapped to acomplex number x+ iy. The original : .s¢ Fourier transform, iginal shape can be recovered { ‘from the wwe “ s(k) bea co-ordinate of a boundary point k , then + s(k)=x(k)+o(k) , om Fourier descriptor: . a(x) =F ESE fourier descriptor of a shape Imaginary axis Xa kr Real axis calculated as follows: The fourier descriptors hi . 1) Find the coor the edge pixels of a shape and put them in a list in order, doing clockwise se the shape. . 2) Define a conte vector using the coordinates obtained. For example, 3) Take the discrete fourier transform of the complex-valued vector. i criptors inherit several properties from the fourier transform. jon invariance: no matter where the shape is located in the image, the rier descriptor remain the sample. ling: If the shape is scaled by a factor, the fourier descriptors are scaled by that factor. 6) Rotation and starting point: Rotating the shape or selecting a different starting point only affects the phase of the descriptors. . Because the discrete fourier transform is invertible, all the lala about the shape is * contained in the fourier descriptors. DIP-105POPULAR PUDLICATIOND. polding? What Is the busi w th 2, What do you moan by global and local thre a? [WBUT 2014] difforonce botwoon ragion growing and aplit morge tech Answer: Part: In yy : vhucre the threshold jon 1 a(n ce 2 Here: be ehonen by the of pray evel theesholdiny, P(r. c) is the feature value at the pixel (r,¢). In P(r) (ree) forall (, ¢) within the imays P(r. c) only, it ix called local threshold. If ¢ depends on sition (r, ec) ax well as onthe feature p(r, c) at that pixel, it is called ylobal threshold. rowing, methods rely mainly on the within one region have si c with its neighbou the cluster as on rtition of an image. It is sometimes call the root of the tree that represents: split into four so they are merged as severa node. This process continues Region growing app Region growing segmentation . The merge 5} s domain, If (depends on the fessture wuimption that the neiyhbouri procedure is lo compare one pi pixels cl be siet to belong, to ified, the pixel id-imerge is Based on a qu ion, This method starts at AP it is found non-uniform then it is f, in contrast, four son squares, are homogene components, The node in the tree is a sepmented ely until no further splits or merges are possible, pposite of the split and merge approach. ide the original images which have clear edges the good id tree 1 optional s scheme, Itrequires a threshold as an input. ¢ of our region growing based segmentation 3. a) Discuss about throshold coding Implomontation. mparativo study of least squaro restoration and constrained least b) a sh [WBUT 2011] A toration mothods. a) Threshold coding is inherently adaptive in the sense that the locatio -ellicients retained for each subimage vary from one subimage to another, In fact, threshold coding is the adaptive transform coding approach most often used in Practise because of its computational simplicity. ‘The underlying concept ist subimage, the transform co-efficients of largest magnitude mak Contribution to reconstructed subimage quality, Bec: efficients vary from on subimage to another, of the transform . for any DIP-106DIGITAL IMAGE PROCESSING rere oF three eas i. threshold a transformed subimage: (1) A single global resold can PP! all subimages, (2) A different threshold can be used for each ima (3) The threshold can be varied as a function of the location of each co- Seient Wi in the subimage, d Minimum meaner error estimation provides a filter that produces optimal oration of a particular image. This may be obtained by the method of least square. In sence of ANY knowledge about noise we may treat it as a measurement 10 yore 4 particular image, we minimize the difference between the ob: age and re-degraded estimated image. mre method of constrained least square can be defined as th rin of some «terion of goodness or quality of image subject to the constraint.that residual norm jetween the image and the re-degraded estimated image be the norm of the noise yectr Constraint least square is a linear image restoration technique but least square petoration is not. 7 Constraint least square retain the simplicity of a spati: ariant linear filter but require prior knowledge of mean and variance of the $ When point-spread function is symmetric, uare restoration does not work operly, to select a solution a constraint Va approach is used. 42) Explain the process of oe n based on segmentation. [WBUT 2016] lates Answer: - In this technique pixels that are to an object are grouped for segmentation. The Thresholding techniques i ind/with region based segmentation. The area that is detected for segmentati e closed. Region based segmentation is also termed as similarity based segment hese would not be any gap due to missing edge pixels in this in this region bi smentation. The boundaries are identified for segmentation. In each and’ every t least one pixel is related to the region and is taken into consideration. identifying the change in the color and texture, the edge flow is converted i ctor. From this the edges are detected for further segmentation. 4) With sketch, explain the object recognition process. [WBUT 2016) Answer: An object recognition system finds object in the real world from an image of the world, \sing object models which are known a priori. This task is surprisingly difficult. Human Perform object recognition effortlessly and instantaneously. Algorithmic description of this task for implementation on machines has been very difficult. object recognition problem can be defined as a labeling problem based on models of Wn objects. Formally, given an image containing one or more objects of interest (and ckground) and a set of labels corresponding to a set of models known to the system, thePOPULAR PUBLICATIONS: : . ions, in the image, system Should assign correct labels to regions, or a set of an without at leas ; object recognition problem is closely tied to the segmentation i without Segmentation partial recognition of objects segmentation can not be done object recognition is not possible. Different components of an object recognition system: * Model database (also called model base) © Feature detector : . ; ; - ents to perform the task An object recognition system Cite) ts to perform * Hypothesizer © Hypothesis verifier | oo The model database contains all the m mn to the system, The information in the qualitative or functional descrip! ecise geometric surface information. The feature detector applies ages and identifies locations of features that help in forming object hypothesis: Pow c) What is pattern fi roach? ) [WBUT 2016) Answi Pattern frit model database depends on the wore for the recognition. It can vary froma i ly scale image. is considered a topographic surface where altitude at In this app : a point is gi its intensity or gray level. The main objective, here, is to fit a pattern over the neighbourhood of the pixel at which edge strength is being calculated. Ps corresponding to the best fit of the pattern are estimated locally. Finally, f edge points are parameters and edges are marked accordingly. ) With appropriate example discuss region’split techni ique for segmentation. b) How isolated point and line can be identified? (WBUT 2018) Answer: 4 a) The basic idea of region splitting is to break the image into a set’of disjoint regions which are coherent within themselves: “ . . . * Initially take the image as a whole to be the area of | interest, * Look at the area of interest and decide if all some similarity constraint. If true then the area of interest.corresponds to a region in the DIP-108 pixels contained in the region satisfy . image.. ronly any neighbouring regions that have identical or si ‘hus, 2 merging Process is used after each split which compare adja ‘ DIGITALIMAs ING If false split the area of interest ( i usually into ider each of the sub-area as the area of interest ‘ ied equal sub-areas) and cons! This process Continues until no fi jurther splitting occ worst case this happened when the areas are just one pixel in sive “ee This is a divide-and-conquer or top-down method a spl ig schedule is used then the final segmentation would “o.. imilar properties. ‘S and merges then if Necessary, algorithms of this nature are called split an: gorithms. qoillustrate the basic principle of these methods let us consid imaginary image. Let I denote the whole image shown in Fig. (i). a : lit asi Not all the pixels in | are similar so the region is Fig. (ii). Assume that all pixels within regions: Ina apes are similar but those in /, are not. Therefore /, is split next as in Fig. Now assume that all pixels within region and that after comparing to be identical. These are thus merged tog: ion are similar with respect to that regions, regions /,, and /4,4 are found in Fig. (iv). Te] bps 1, betes 1 elt safle aa (i) Whol (ii) First split (iii) Second split (iv) Merge i Fig. Example of region splitting and merging dA the most basic type of discontinuity in a digital image. The most commor appr ‘0 finding discontinuity is to run an (nxn) mask over each point in the image . The mask is shown below. (A mask for point detection)POPULAR PUBLICATIONS oo. mask is centered. If the The point is detected at 2 location (x. y) in an image where the corresponding value of R such that | 8] > T and T is non-negative . one i image > Where R is the response of the mask at any point in the image end’ 8 Mt (xy threshold value. It means that isolated point is detected at the corresponding ra e Mn This formulation serves to measures the weighted differenc ie that and its neighbours since the gray level of an isolated point will be very of its neighbors. _ ontinuity. For direction the y : h ent Point of a line is most associated with. For line detection, mask, ith end jth mask. Then we have, Rl > lal i “ » It means that the corresponding points is more likely eased wit a ine athe Line detection is the next level of complexity in the direction of i 2ny point in the image, a response can be calculated that will Girection of the mask i. Line detection masks in Horizontal direction (Fig. a) 45° direction (fig. (b)). vertical direction (fig. c) and 45° direction (fig. (d)) is shown below. =1{-1]2 aT 2]1 2 |-1]-1 Fig(®) ig. (c) Fig.(d) The greatest iculation from these matrices will yield the direction of the given IWBUT 2003] [WBUT 2011] G) Edge detecting operations Hae ao an ©) Split and Merge Algorithm ll Answer: QWeUT 2014] a) Optimum Thresholding: . Thresholding is a non-linear Operation that converts a gray-scale image i image where the two levels are assigned to pixels that are below 7 atone. Ln : — threshold value we can apply a threshold to data directly from the command lii Specified €.g. my Binary Image = my Gray Image > Threshold Value 255:0 menis however for more efficient to use the image threshold operation which also provides geveral methods for finding the “optimal” threshold value for a given image. Image . threshold provides the following methods for determining the threshold value. |, Automatically calculate a threshold value using an iterative method. 2. ‘Approximately the histogram of the image as a bi-modal distribution and choose a mid-point value as the threshold level. . 3. , Fuzzy thresholding using as the measure of Fuzziness. : 4, Evaluate the threshold based on the last 8 pixel in each row, usi rating rows (Adaptive. thresholding), : . b) Constrained least squares: . The constrained least square filier is another approach fi ove ing some of the difficulties of the inverse filter (excessive noise amplification) and of the weiner filter (estimation of the power spectrum of the ideal image), white still retaining the simplicity _ of a spatially invariant linear filter. But it is required to have a prior knowledge about mean and variance of the noise. The constraint le: re algorithm is based on finding adirect solution using a criterion C , which ensures optimal smoothness of the deblurred image. It is a linear image restoration techni ich the smoothness of the restored image is maximized subject to a constrai fidelity of the restored image. ¢) Contrast stretching: Contrast stretching is a simple im: ancement technique that attempts to improve the desired range of values e. ¢ full range of pixel values that the image type concerned allows. It differs fror sophisticated histogram equalization in that it can only apply a linear scaling function to the image pixel values. As a resillt the enhancement is less harash. . . contrast in an image by het he range of intensity values it contains to span a d) Edge det g operations: : 4 Edge deté ‘i an image processing technique for finding the boundaries of objects i 5. It works by detecting discontinuities in brightness. Edge detection is used for ii segmentation and data’extracting in areas such as image processing, computer vision and machine vision. Common edge detection algorithms include sobel, canny, Prewitt, Roberts and fuzzy logic methods. €) Split and Merge Algorithm: : , , ‘ * Splitting and’ Merging attempts to divide an image into uniform regions. The basic Tepresentational structure is pyramidal i.e. a square region of size m by m at one level of .2 pyramid has 4 sub-regions of size m/2by m/2 below it in the pyramid, Usually the DIP-111POPULAR PUBLICATIONS algorithm starts from the initial assumption that the entire image is a single region, then computes the homogeneity criterion to s¢e if itis TRUE, If false, then the square region is split into the four smaller regions. This process is then repeated on each of the sub- regions until no further split ig is necessary. These small square regions are t! lerged if they are similar to give larger irregular regions. The process terminates WI her merges are possible.\UESTION 2014 . GROUP-A (Multiple Choice Type Questions) 4, choose the correct alternatives for any ten of the following: i) Golour image processing is gaining importance because a) It's more pleasant to watch b) It's cost effective ¢) It's easy to Capture and represent /¢) Use of digital image over the internet has increased signifi v . + jy Wavelets are \ va) Foundation for representing image in various degree: if resolution. b) Of remarkable advantage for image compression’ ) Both (a) and (b) d) None of the above iil) Representation & description almost always output of a Ya) Segmentation stage b) Filtering stage c) Compression stage 4) All of the above iv) Coloured Model Names: : ; a) The RGB colour mo b) The CMY & CMYK colour models c) The HSI colour model d) (a) & (b) only Ve) (2), (c) & (0) ; v) The basic prit of compression matches the principle of: a) Channel b) Line coding c) Source coding, Vd) All of these vi) Which is used for image-compression: transfer coding b) Wavelet coding coding . ¥d) Convolution coding vii) Not a morphological image procession algorithm: a) Thinning : b) Skeleton %c) Both (a) & (b) 4) None of these viii) The basic principle of compression matches the principle of a) Channel coding oa b) Line Coding ¢) Source Coding ~ Vd) All of theseOe ck unnafet Ct b) hulimen eee isfer coding 7 SOoveodeg : ) Wavelet coding ated point can be best detected using 2 * ry " detivalve . vb) 2" derivative ©) 3" derivative d) None of these xi) In th will of course be. b) Higher intensity Value . %a) Lower intensity value . c) Medium intensity value ) None of these led xii) Measuring an intensity value of a fixed pixel including the effect of the neigh! a) Averaging b) Spatial filtering ¥c) (a) &(b) both ) None of these 7 . . GROUP-B ww ° ns. intensity distribution scale the background (Short Answer Type Questio 2. What is Weber Ratio? Show the variation of Weber ratio for hu . er Type Question No. 11. Sce Topic: DIGITAL IMAGE PROCESSING SYSTEMS, Sho! metrical transformation? Develop the 3. What are the basic steps involved in image homogeneous from of this transformation? Sce Topic: DIGITAL IMAGE me 1S, Short Answer Type Question No. 12. fat is energy compaction of image Transform? 4, State the applications of image transf¢ iS JENTATION), Short Answer Type Question No. 10. See Topic: IMAGE TRANSFORM: 5. What is Holomorphic Filtering? Setup the equation for this filter. See Topic: IMAGE E| IN THE SPATIAL AND FREQUENCY DOMAIN, Short Answer Type Questi 6. What is bounds tion in Morphological image operation? See Topic: MORPHOLOGICAL IMAGE PROCESSING, Short Answer Type Question No. 1. ~ @ GRour-c (Long Answer Type Questions) 7. a) Explain unsharp masking and high boost fitering. b) Write down the laplacian operator for an image of two variables. a Mat is unitary transform? Define and compute the equations for unita ransform, inverse Fourier transform for both 1-D and 2-D image eee our : / DIP-114DIGITAL IMAGE PROCESSING ») & b) See Topic: IMAGE ENHA NCI ‘ short Answer Type Question No, 7, EMENT IN THE SPATIAL AND FREQUENCY DOMAIN, _ g)see Topic: IMAGE TRANSFORMS (MPLEMENTATION) Long Answer Type Question No. 6. . a) Explain the operation of a Median fitter. p) Describe Bilinear interpolation method, ¢) What is Hough transform and where is it used? ) See Topic: IMAGE ENHANCEMEN’ answer Type Question No, 8, ahead satan _ nse To maces, IMAGE PROCESSING SYSTEMS, Short Answer Type see Top RANSFORMS (IMPLEMENTATION), Long = sw 9. a) Write down the usefulness of segmentation? \ b) Briefly explain about region 9rowing technique. ° ¢) Explain the role of discrete cosine transform in image process! Question No. 4(b). 1) See Topic: IMAGE SEGMENTATION, REPRESENTAT| DESCRIPTION, Short Answer Type Question No. 2. b) See Topic: IMAGE SEGMENTATION, REPRES IN’ 10N AND DESCRIPTION, Short Answer Type Question No. I(or). «) Ste Topic: IMAGE TRANSFORMS (IMP! 10. a) What do you mean by redundancy? Wi b) What is compression ratio? Consider it image of 256256 pixels. After compression, the image size is 6,554 bytes. Find the cx sion ratio. ¢) Describe image compressi tem. See Topic: IMAGE DATA Ct ESSION, Long Answer Type Question No. S(a), (b) & (¢). 11. a) Explain the resto odel for continuous function in detail, ») What is the role uantization in image processing? ) What is the differé tween lossy and lossless compression? @) Whatis salt and er Noise? What is Gaussian noise? + a) See Topic: Or ENHANCEMENT IN THE SPATIAL AND FREQUENCY DOMAIN, Shor jon No. 14, AGE DATA COMPRESSION, Short Answer Type Question No. 3. : IMAGE DATA COMPRESSION, Short Answer Type Question No. 2. 4) See Topic: IMAGE ENHANCEMENT IN THE SPATIAL AND FREQUENCY DOMAIN, Shor Answer Type Question No. 13. 12. Write short notes on any three of following: @) Spatial filtering b) Edge detecting operations ¢) Brightness adaptation 9) Walsh Transform €) Hit-or-miss Transform f) Filtering in frequency domainJoe g) Split and Merge Algorithm. . le a) See Topic: IMAGE ENHANCEMENT IN THE SPATIAL AND FREQUENCY DOMAIN, Long Answer Type Question No. 11(e). s b) See Topic: IMAGE SEGMENTATION, REPRESENTATION AND DESCRIPTION, Long Answer __ Type Question No. 6(d). r * ©) See Topic: DIGITAL IMAGE PROCESSING SYSTEMS, Long An: 4) See Topic: IMAGE TRANSFORMS (IMPLEMENTATION), Long Answer Type ¢) See Topic: MORPHOLOGICAL IMAGE PROCESSING, Long Answer Type Quest! f) See Topic: IMAGE ENHANCEMENT IN THE SPATIAL AND FREQUE! Answer Type Question No. 1 g) See Topie: IMAGE SEGMENTATION, REPRESENTATION we IN, Long Answer ‘Type Question No. 6(e). TION 20: . . GROUP ~A (Multiple Choice Type Q 1. Answer any fen questions: i) The computation of Walsh coefficient ir a) only subtraction b) only addition Vc) addition and subtraction d) none of the above ii) Compression is achieved. a) filtering . b) multiplying an array ¥c) minimizing the * — d)none of these iil) Which of the ing is not an image file format? a) TIFF )) BMP. °) GIF vd) none of these iv) The trai which is widely used in JPEG compression scheme is 7 bIDFT r rd Transform Yd) Discrete Cosine Transform vy) let-based image-compression scheme overcomes the problem of a) Blur object -%b) Blocking artifact c) Background sharpening d) all of the abovevi) The smallest units of a digital image is represented by a) Aone dimensional matrix b) Logic 0 or 4 ¢) Dot ¥ 4) Pixel vii) Which of the following statement is true? a) The resolution of CMOS image sensor is better than CCD image : Yb) The resolution of CCD image sensor is better than CMOS image Dm c) CCD image sensor is cost effective t it 4d) none of the above menos iiece ix) What operation is needed for background removal? a) Image multiplication b) Image addi . ¢) Image subtraction . Yd) all of 7 ‘ x) Wavelets are a) of remarkable advantage for image compression 7b) foundation for representing images in vario resolution ) both (a) and (b) d) none of these xi) An example of volume image is a) Aone dimensional image is b).A two dimensional image * ¥c) A three dimensional = d) all of the above xi) Basic principle of compression matches the principle of a) Channel coding b) Source coding ¢) Line coding 4) none of these GROUP-B (Short Answer Type Questions) ariables. IMAGE ENHANCEMENT IN THE SPATIAL AND FREQUENCY DOMAIN, Short Question No. 7. 2. Explain fio asking and High boost filtering. Write the expression of Laplacian operator for an image o' a 7 ‘ ‘ 3. If an image is rotated by an angle 7 will there be any change in the histogram of that image’ Justify your answer. D See Topic: IMAGE TRANSFORMS (IMPLEMENTATION), Short Answer Type Question No. 11. 4. Explain bilinear interpolation method used for image zooming. See Topic: DIGITAL IMAGE PROCESSING SYSTEMS, Short Answer Type Question No. 5. DIP-117POPULAR PUBLICATIONS 5. Explain the operation of an ord See Topic: MORPHOLOGICAL (MAGE PnocEsNine toeta IMAGE PROCESSING, Short Answer Type Quéstion No. 2- sheared anes about the usefulness of discrete cosine transform. 2 E TRANSFORMS (IMPLEMENTATION), Short Answer Type ROS GROUP-C (Long Answer Type Questions) specification? by What is the difference between spatial domain filtering and freq edge detection operator and contrast stretching of an image. ©) Briefly describe the smoothing linear spatial filtering. What is bit plane slicing method? a) See Topic: IMAGE TRANSFORMS (IMPLEMENTATION), Short Answer Type Question No. 42, b) & c) See Topic: IMAGE ENHANCEMENT IN THE SPATH 1D FREQUENCY DOMAIN, Long Answer Type Question No. 7. 7. a) What is Histogram of an-image and why processing? What do Weceerencee 8. a) Whatis m-connectivity among pixels? Give b) What do you mean by global and | region growing and split merge technique? c) What is Hough transform and where i sholding? What is the basic difference between SYSTEMS, Short Answer Type Question No. 13. b) See Topic ‘THE SPATIAL AND FREQUENCY DOMAIN, Long - Answer Type Question No. 6, ©) See Topic: IMAGE TRA! (IMPLEMENTATION), Long Answer’ Type Question No. 4(b). 9. a) Why image comy jonis required? - b) Take any 4x: e of your choice and show how block truncation coding can be applied to compress it. v . ¢) Distinguish " image enhancement and image restoration. IMAGE DATA COMPRESSION, Long Answer Type Question No. 6(a) & (b). AGE ENHANCEMENT IN THE SPATIAL AND FREQUENCY DOMAIN, Short ¢ Question No. 2. 40. a) Classify different type of redundancies in an image. Mention the different video compression techniques and briefly discuss about any one of them. b) Discuss’about spatial resolution and gray — level resolution c) What is Homomorphic filtering? Mention the domain of application of this filter. . a) See Topic: IMAGE DATA COMPRESSION, Long Answer Type Question No. 6(c). b) & c) See Topic: IMAGE ENHANCEMENT IN THE SPATIAL AND FREQUENCY DOMAI! ano Ancwer Tyne Question No. 8.11. Write short notes on any three of the following: @) Local enhancement . b) Vector quantization method for ‘image compression ©) Piecewise Linear transformation ) Erosion and Dilation ©) Image sharpening filters. 3) See Topic: IMAGE ENHANCEMENT IN THE SPATIAL AND FREQUENCY Long Answer Typé Question No. 11(). b) See Topic: IMAGE DATA COMPRESSION, Long Answer Type Question No. MAGE TRANSFORMS (IMPLEMENTATION), Long Answer Type Question No. 7(f) UENCY MORPHOLOGICAL IMAGE PROCESSING, Long Answer T: jon No. 4(c). IMAGE ENHANCEMENT IN THE SPATIAL AND DOMAIN, Long Answer Type Question No. 11(g). \UESTION 201 Group-A (Multiple Choice ‘Type . 1. Choose the correct alternatives for any ten of 19: ‘. . i) The spectrum of the visible light is a) 10-350 nm ¥b) 380-760 nm c) 760 nm and above 4d) none of these il) The image acquisition process is u: a) store image b) transform image ¢) display image . Yd) none of these iii) Colour image is nted by © a) 2-bit b) 8-bit Vc) 24-bit d) 64-bit iv) Which of the following operations is idempotent? a) dilatic b) erosion ¢) convolution Yd) dosing wi T of { (m,n) > F(k,D} then forthe sequence f(m— m,n) the DFT becomes (108) FC 1) by eh" BCR LY . “o) e248 ECD) d) eU2*" ECK 1 vi) The kernel for 2D-DFT with N = 4 is calculated by the command a) dft2(4) Yb) fft2(4) ©) dfimtx(4) 4) none of these DIP-119POPULAR PUBLICATIONS vil) DCT is widely used for a) image degradation ~b) image compression ¢) image restoration 9) none of these Vill) The photosensitive detector of the human eye is a) eye lens b) iris _ ¥e)retina 9) ix) Through decimated by 2 operations the sampling rate is a) increased Yb) decreased ©) none of these x) The histogram equalization process a) blurs the image b) fades the im Yc) improves the brightness of the image d) none of - xi) Ifthe size of the mask for averaging is increased, the image will be a) noise free Yb) blurred ©) degraded d) none of these xii) Salt and pepper noise can be removed by a) weighted average filter b) Gaussian fiter ~c) Median filter igh boost filter xiv) Edge detection of an image broa a) low spatial frequen ©) thresholding low xii) An example of dictionary based codit is . = . a) Huffman, b) run-length c) Lempel-Ziv-Weich )-predictive ans hancement b) high frequency’enhancement + ¥d) detection of intensity variation xv) The classical H ansform is concerned with the identification of Ya) lines in * = b) zeros in an image C) poles in an image d) boundary in an image xvi) A ve nstrm is a special case of ransforms b) Z-transform c) Fourier transform 4) discrete cosine transform : Group-B (Short Answer Type Questions) 2. a) Differentiate between image and scene. b) What are trie image sensors and how are they used? Sce Topic: DIGITAL IMAGE PROCESSING SYSTEMS, Short Answer Type Question No. 14. DIP-120DIGITAL IMAGE PROCESSING do you understand by the terms ‘hi "and" on 3.2) What terms ‘histogram’ and ‘histagram equalization’? What are the steps to perform histogram exusization? iis ¢) What is the necessity for histogram equalization? aysee Topic: IMAGE TRANSFORMS (IMPLEMENTATION), Long Answer Type Question No. 4a). p) & «See Topic: IMAGE TRANSFORMS (IMPLEMENTATION), Short Answer Type Question No- 13¢a) & (b)- 4,a) What do you mean by aliasing in the context of image sampling? Explain. p) What do you mean by the term ‘image file format’? Mention some of the used image file format. see Topic: DIGITAL IMAGE PROCESSING SYSTEMS, Short Answe Ty 5, a) Briefly explain skeletons. NN b) Write down the effects of the dilation and erosion a process. __ See Topic: MORPHOLOGICAL IMAGE PROCESSING, Short Answer Type Question No- 3- 6 Show how Prewitt and Sobel operators can be us th ie noise. see Topic: IMAGE ENNANCEMENT IN THE Re AND FREQUENCY DOMAI ‘Answer Type Question No, 17, : c . nswer Type Questions) : the components of image processing system. Je enhancement using arithmetic operator. . ion No. 15. Short ”. 7.a) With neat sketch explain brie! b) Explain the process of ir : 22 c) What is-the need f ation technique? Consider the image F -(3 +}: Calculate the |AGE PROCESSING SYSTEMS, Short Answer Type Question No.9. HANCEMENT IN ‘THE SPATIAL AND FREQUENCY DOMAIN jit colour image? For what purpose could it be used? Explain. in the model of image restoration process in presence of noise. jin the process of region extraction based on segmentation. ic: DIGITAL IMAGE PROCESSING SYSTEMS, Short Answer Type Question No. 10. - + IMAGE ENHANCEMENT IN THE SPATIAL AND FREQUENCY DOMAIN, Lor Answer Type Question No. 9(¢). ©) See Topic: IMAGE SEGMENTATION, REPRESENTATION AND DESCRIPTION, Long Answ Type Question No. 4(a). 2 ie 9. a) With neat sketch, explain the object recognition process. b) Explain the lines and edges detecting process with mathematical models. ‘C) What is pattern fritting approach? DIP-1212) & ©) See Topic: IMAGE SEGMENTATION, REPRESENTATION AND DESCRIPTION, Long Answer Type Question No. 4(b) & (c). 'b)Sce Topic: MORPHOLOGICAL IMAGE PROCESSING, Short Answer Type Question No. 4. 10. a) Whats the need for compression? What are different compression methods? b) Discuss briefly the concept of Huffman coding. ¢) Define interpixel redundancy? ° : ‘See Topic: IMAGE DATA COMPRESSION, Long Answer Type Question No. 7(2 eee 11. Write short notes on any three of the following: a) Walsh Transformation b) Image Restoration mode! NN ¢) Hotelling Transformation “d) Run length coding e) Chain code . : ic: IMAGE TRANSFORMS (IMPLEMENTATION), Long Answer Type Question No. 7(d). 'b) See Topic: IMAGE ENHANCEMENT IN THE SPA’ FREQUENCY DOMAIN, Long TION), Long Answer Type Question,No. 7(g). 4) See Topic: IMAGE DATA COMPRESSION, Lo + Type Question No. 8(g). IG, Long Answer Type Question No. 4(d). Group-A (Multiple Choice Type Questions) s for any fen of the following: ctive brightness for human? : glare limit b) photopic threshold to glare limit a) scotopic thy ¢) scotopic threshold to infinity . ) photopic threshold to infinity ii) Repres enaton and description almost always follow the output of a 4 ;ntation stage ‘ ») fittering stage ¢ ression stage ~ ¢) all of these iil) An image is a 2D array of Ya) digital data b) electrical signals ¢) photographic objects 6) light signalsiv) The image function f(x, y) is characterized by two components: S(%Y)= (XY), y) where nd a) 0 is T :G.compre: a) FFT ; by Ii ¢) Hadamard transform discrete.cosine transform ix) Which of the following transforms is used for line detéction in image processing? a) hadamard vb) haugh ¢) haar d) slant x) Which of the following image processing transforms does not satisfy the separability property? a)Walsh ‘ourier - ¢-) DCT Yd) Hotelling xi) Which of the followi rai ot an image file format? . a) TIFF IP . ¢) GIF vd) none of these xii) Which one of vetoes is the lossy image compression? Ya) TIFF b) BMP 7 ¢) GIF d) none of these Group -B (Short Answer Type Questions) 2. What is Weber ratio? Show the variation of Weber ratio for human eye. See Topic: DIGITAL IMAGE PROCESSING SYSTEMS, Short Answer Type Question No. 11. 3. If an image is rotated by an angle of 60° will there be any change in the histogram of that image? Justify your answer. See Topic: IMAGE TRANSFORMS (IMPLEMENTATION), Short Answer Type Question No. 11. DIP-123POPULAR PUBLICATIONS 4. Explain the operation of a median filter, . See Topic: IMAGE ENHANCEMENT IN THE SPATIAL AND FREQUENCY DOMAIN, Short Answer Type Question No. 8, 5. Distinguish between image enhancement and image restoration. ‘See Topic: IMAGE ENHANCEMENT IN THE SPATIAL AND FREQUENCY DOMAIN, Answer Type Question No, 2(I" Part), 6. What is boundary extraction in Morphological image operation? See Topic: MORPHOLOGICAL IMAGE PROCESSING, Short Answer Type Qu Na ‘ . . Group—C ‘ (Long Answer Type Questions) A 7. a) Consider the image segment 1 4a20Nn 1 2 1 2 1. 0 . - ; Let’ v=(0,1}. Compute the (i) shortest 4-ad) ‘and (jl) m-adjacency distance between the pixels p and q._ ; See Topic: IMAGE SEGMENTATION, REPRI ION AND DESCRIPTION, Short Answer ‘Type Question No. 3. . b) Discuss one technique of image acqui See Topic: DIGITAL IMAGE ESSING SYSTEMS, Short Answer Type Question No. 16(a). ¢) Write down the key See Topic: DIGITAL I! processing and explain them. ING SYSTEMS, Short Answer Type Question No. 9. 8. a) What do,you ge negative? Expiain. ) Explain Intensity slicing with example. ) Why do we need Log Transformation in dynamic range compression? See Topic: ENHANCEMENT IN THE SPATIAL AND FREQUENCY DOMAIN, Long Answe No. 3(a), (b) & (¢). the usefulness of segmentation. , See Topic: IMAGE SEGMENTATION, REPRESENTATION AND DESCRIPTION, Short Answer ‘Type Question No. 2. i ») Describe the region growing technique for image segmentation and mention the problem associated with it. ' IMAGE SEGMENTATION, REPRESENTATION AND DESCRIPTION, ‘Type Question No. 1. : Short Answer DIP-124DIGITAL IMAGE PROCESSING ¢) Discuss the Hough transform method for ed ¢-linking. see TOP! MAGE TRANSFORMS (IMPLEME ° ENTATION), Long Answer Type Question No. 4(b)- 40. a) What do you mean by image re nition? bp) What is meant by classification of iene ¢) Describe any image recognition technique. a) & ¢) See Topic: MORPHOLOGICAL IMAGE PROCESSING, Long Answer tion No. 34a) & (b). f b) See Topic: DIGITAL IMAGE PROCESSING SYSTEMS, Short Answer Ty; No. 16(b). 11, Write short notes on any three of the following: : a) Edge detection operation NN b) Hit-or-miss transformation . c) Smoothing linear spatial filtering . 4) Brightness adaptation €) Image compression techniques. »S a) See Topic: IMAGE ENHANCEMENT IN THE SP, Answer Type Question No. 11(i), b) See Topic: MORPHOLOGICAL IMAGE PR , Long Answer Type Question No. 4(b): «) See Topic: IMAGE ENHANCEMENT 1 ATIAL AND FREQUENCY DOMAIN, Long Answer Type Question No. 11(j). - : DIGITAL IMAGE PROC SYSTEMS, Long Answer Type Question No. 2(b)- IMAGE DATA COM! ng Answer Type Question No. 8(h). QUESTION 2018 ND FREQUENCY DOMAIN, Long “Group-A (Multiple Choice Type Questions) 1. Choose the cor alternatives for any ten of the following: ’) Huffman coding approach reduces va) ;dundancy only b) inter-pixel redundancy only \g and inter-pixel redundancy 4d) psycho-visual redundancy only ii) The total amount of energy that flows from the light source and it is usually measured in watt W) is called , : a) radiance %b) luminance ©) reflectance 4) none’of these iil) The effect caused by the use of an insufficient number of gray levels in smooth areas of a digi image is called Ya) false contouring) gray level slicing C) bit plane ¢) thinningPOPULAR PUBLICATIONS iv) In 8 distance measurement system distance between centre pixel and a comet pixel is Ya) 2 units ») V2 units “* ©) 1 unit d) 1.5 units v) Edge detection of an image broadly means * a) low spatial frequency enhancement b) high spatial frequency enhancement ©) thresholding low spatial frequencies. ¥d) none of these vi) The classical Hough transform is concerned with the identification of Dm Ya) lines in an image b) zeros in an image ¢) poles in an image d) none of these vi) The basic principle of compression matches the principle of N a) channel coding) line coding ~ ¢) source ae +) eh ofthese viii) A Wavelet transform is a special’case of aaan @) Laplace transform b) Z-transform 7) Fourier transform. d) none of th tness ix) Averaging filter is used for . \ a) sharpening b) contrast nt x) Wiener filter is used for ~ a) restoration b) “> ) sharpening ¥d) none of these vd) smoothing xi) Ifa function f(x, y) is finite in domain, the Fourier transform of f(x, y) will be Ya) finite infinite ¢) undefined d) zero all the coefficients are equal is called b) median filter Yc) box ae : d) none of these Group-B @ (Short Answer Type Questions) the basic steps involved in the image geometrical transformation? Develop the \e0us form of this transformation. Sce Topic: DIGITAL IMAGE PROCESSING SYSTEMS, Short Answer Type Quiestion No. 12. 3. What is Homomorphic Filtering? Set up the equation for this filter. See Topic: IMAGE ENHANCEMENT IN THE SPATIAL AND FREQUENCY DOMAIN, Short Answer Type Question No. 16. 4. Brightness discrimination is poor at low levels Of illumination. Explain. See Topic: DIGITAL IMAGE PROCESSING SYSTEMS, Short Answer Type Question No. 17. DIP-126 7