Image Processing November 2004

Time : 3 Hrs.] N.B.: [Marks : 100

(1) Question No. 1 is compulsory. (2) Attempt any four out of remaining six questions. (3) Figures to the right indicate full marks. (4) Assume suitable data if necessary.
[20]

1.

Justify / contradict following statements : (a) Enhancement process does not change the information content of image. (b) For digital image having salt pepper noise, median filter is the best filter. (c) For continuous image histogram can be perfectly equalized, but it may not be so for digital image. (d) The statement that Run length coding gives data compression is not always true. (e) Quality of picture depends on the number of pixels and gray level that represent the picture.

2.

Let 0(n), 1(n), 2(n) . . . . N 1 (n) be a set of N orthogonal sequences each of length N. (a) Show that N2 orthogonal pattern of size N N can be generated by using these sequences. Prove the orthogonality of these patterns. [6] (b) Write first four Walsh sequence of length N = 4, using these sequences generate sixteen orthogonal Walsh pattern. [8] (c) For the following 4 4 image show that any four patterns from the above sixteen pattern are required to fully represent the image. Identify these pattern : [6] 4 4 10 10 4 4 10 10 8 8 12 12 8 8 12 12 [8]

3.

(a) What is the impulse response of each filter ? (i) Transfer function is : H1(z1, z2) = 1 a1z 1 a2z 1 a3 z1 1z2 1 a4 z1, z2 1. (ii) Frequency response is H(w1, w2) = 1 2 cos w1 2 cos w2. (b) (i) For the 2 2 transform A and the image U : A=

[8]

1 2

3 1

1 3

2 3 U= 1 2
[6]
N 1 u1

Calculate the transformed image v, and the basis image. (ii) Prove two dimensional energy conservation relation.
N 1 x1

| f (x, y) | 2 =

y=0

| F(u, v) | 2

v =0

where f(x, y) represent 2 D image signal and F(u,v) represents transformed domain signal. 4. (a) Give below is the table of eight symbol and their frequency occurrence. Symbol Frequency s0 0.25 s1 0.15 s2 0.06 s3 0.08 s4 0.21 s5 0.14 s6 0.07 s7 0.04 [4] [4] [4] [8]

Give Huffman code for each eight symbols. (b) (i) Evaluate minimum number of average bits sequence per symbol. (ii) What is coding efficiency for the code you have obtain in 4(a). (iii) What is sumlength coding ? Explain clearly. 5.

(a) Explain in detail enhancement techniques in special domain used for images. [10] (b) (i) Discuss thinning algorithms for finding the skeleton of a region. Apply it to a square and a triangle.[6] (ii) Illustrate opening and closing operations with suitable example. [4] (a) Write an expression for a two dimensional DFT. What is its relationship with one dimensional DFT ? [6] (b) (i) Show that onedimensional FFT algorithm can be used to compute two dimensional DFT of a digital image. [7] (ii) Using 4 point FFT algorithms evaluate 20 DFT of following image. [7] 0 1 2 1 1 0 1 2 2 1 0 1 1 2 1 0

6.

University Question Papers

7.

Write short note on following : (a) Chain code. (c) Hit or miss transform. (e) Dilation and Erosion.

[20] (b) Fourier descriptor. (d) Uniform sampling and non uniform sampling.

Image Processing May 2005

Time : 3 Hrs.] N.B.: [Marks : 100

(1) Attempt any five questions. (2) Figures to the right indicate full marks. (3) Assume suitable data wherever required but justify the same.

1.

(a) Write DDA Algorithm to draw a line, calculate all points using the same algorithm to draw line between points x(10, 10) and y(20, 15). [10] (b) Show that the first order difference of a chain code normalize it to rotation. Compute the first difference of code 0101030303323232212111. [10] (a) Define : (i) Eucledean distance (ii) City block distance (iii) Chess board distance Consider the following subimage; determine 1 2 3 4 5 6 7 8 1 (i) Eucledean distance 2 q (ii) City block distance 3 (iii) Chessboard distance between p and q 4 5 6 7 8 9 10 p (b) Explain with suitable example region splitting and merging technique for image segmentation. [10]

2.

[10] [10]

3.

(a) For the 2 2 transform A and the image U A=

1 1 1 , 2 1 1

1 2 U= 3 4
[10] [10]

Calculate the transformed image V and the basis image. (b) Explain with suitable example morphological filter to eliminate background noise. 4. (a) Define the following spatial filters (i) Median (ii) Min. (iii) Max. Consider the digital image 0 1 0 6 7 2 0 1 6 5 1 1 7 5 6 1 0 6 6 5 2 5 6 7 6 Calculate the value at point g(2, 2) = 7 for above filters. (b) Draw and explain block diagram of JPEG encoder and decoder. 5.

[10]

(a) Show that a high pass filtered image can be obtained in spatial domain as Highpass = Original Lowpass. [10] (b) Suppose that a digital image is subjected to histogram equalization. Show that a second pass of histogram equalization will produce exactly the same result as the first pass. [10] (a) Draw and explain the 2D fast wavelet transform filterbank. (b) Gray level histogram of an image is given below : Gray level Frequency 0 0 1 0 2 26 3 39 4 16 5 84 6 20 7 0 [10] [10]

6.

Compute the gray level histogram of the output image obtained by enhancing the input by histogram equalization technique.

Vidyalankar : BE IP

7.

Write short notes any four : (i) Image sampling (ii) Walsh transform (iii) Psychovisual redundancy. (iv) Fourier Descriptor. (v) Bit plane coding (vi) Butterworth filter.

[20]

Image Processing December 2005

Time : 3 Hrs.] N.B.: (1) (2) (3) (4) Question No. 1 is compulsory Attempt any four out of remaining six questions. Figures to the right indicate full marks. Assume suitable data if necessary. [Marks : 100

1.

(a) Write DDA Algorithm to draw a line. Also explain the basic transformations in compute Graphics. [10] (b) Chain codes can be made invariant to translation and rotation. Explain. Compute the first difference of code 011103022232212111. [10] (a) Develop an algorithm for converting one pixel thick 8 connected path to 4connected path. (b) Suppose, a 8 8 pixel image represented by 4 bits/pixel has the following grey level distribution Grey level (nx) No. of pixel P(nx) Grey level (ng) No. of pixel P(ny) 0 4 0 24 1/7 16 1/7 6 2/7 17 2/7 8 3/7 11 3/7 10 4/7 8 4/7 12 5/7 4 5/7 11 6/7 4 6/7 9 1 0 1 6 [10] [10]

2.

Modify above histogram such that the desired distribution is as follower,

3.

(a) Obtain Hadamard matrix H(8) from a simple second order matrix H(2), and draw flow diagram for H(8). [10] (i) Define Zequency and find it for H(8). (ii) Obtain the watch matrix w(8) from H(8). (b) Explain the basic principle of transform coding for image compression and illustrate same with the help of DFT and DCT. [10] (a) Explain dilation, erosion, opening and closing morphological operations. Also erode the following region of 1s. Using a 3 3 square operator. [10] 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 1 1 1 (b) Explain in detail enhancement techniques in spatial domain used for images.

4.

[10]

5.

(a) Explain with suitable example region splitting and merging technique for image segmentation. [10] (b) Write an expression for a two dimensional DCT. What is its relationship with one dimensional DCT ? Find DCT for the given image. [10] 2 1 1 2 0 1 0 1 1 0 0 2 0 1 1 3 [10]

6.

(a) Explain the filter in spatial domain (i) Low pass filter (iv) Butterworth filter (ii) High pass filter (v) Gaussian filter. (iii) Median filter (b) Explain in detail, what is image pyramid and subband coding. Write short notes on following (a) Hit or Miss transform (b) Polygonal approximation and signature. (c) Huffman coding and Run length coding. (d) Hough Transform.

[10] [20]

7.

University Question Papers

Image Processing May 2006

Time : 3 Hrs.] N.B.: (1) Attempt any five questions out of seven questions. (2) Figures to the right indicate full marks. (3) Assume suitable data if necessary. [Marks : 100

1.

(a) Write DDA algorithm to draw a line, calculate all points using the same algorithm to draw line between points x(10, 10) and y(20, 20). [10] (b) Explain 2D bilevel image compression standard with flowchart. [10] (a) Show that subtracting the Laplacian from an image is proportional to unsharp masking. (b) Compute a two scale fast wavelet transform of sequence f(n) = {1, 2, 3, 4} [6] [9]

2.

1 n = 0,1 Given : h (n) = 2 0 otherwise 1 n=0 2 1 h (n) = n =1 2 0 otherwise

(c) Justify Hough transform used to detect the curves. 3. (a) Define erosion and dialetion. What is net effect of opening and closing using gray scaled image? (b) Apply KL transform on Given x1 = (0, 0, 0)T, x2 = (1, 0, 0)T, x3 = (1, 1, 0)T and x4 = (1, 0, 1, )T. (a) Gray level histogram of an image is given below : 0 1 2 3 4 5 6 Gray level 123 78 281 417 639 1054 816 Frequency Compute histogram equalization. Draw the histogram. (b) Obtain the slant transform matrix for N = 8. Write down the properties of slant transform. 5. (a) What is pruning? Explain the different operation required. (b) Compare and contrast : (i) ideal HPF Vs ideal LPF. (ii) Gaussian HPF Vs Gaussian LPF. (c) Given a 44 image whose gray levels, order lexicographically are as follows : 3120203123101302 Calculate the spatial moments upto second order. (a) Explain in details Homomorphic filter and morphological filters. (b) Show that Highpass = Orgional Lowpass. Write short notes on : (a) Edge Linking and Boundary detection via graph theoretic Technique. (b) skeleton using morphological algorithm. (c) DPCM/DCT encoder for video compression (d) Subband coding. 7 688 [12] [5] [6] [5] [10] [10] [8]

4.

[9]

6. 7.

[12] [8] [20]

Image Processing November 2006

Time : 3 Hrs.] N.B.: (1) Question No. 1 is compulsory. (2) Attempt any four questions out of remaining six questions. (3) Assume suitable data if necessary. [15] [Marks : 100

1.

(a) Explain following terms: (i) Isopreference curve (ii) Mach band (iii) Simultaneous contrast (iv) Uniform and Nonuniform sampling (v) Fourier descriptor
7

Vidyalankar : BE IP

(b) Consider the two image subset S1 and S2 shown in the following figure for V = {1} determine whether these two subset are [5] (1) 4adjacent (2) 8adjacent (3) Madjacent S1 S2 0 1 1 0 0 2. 0 0 0 0 0 0 0 0 1 1 0 1 1 1 1 0 0 0 1 1 0 0 1 0 0 0 1 1 0 0 1 0 0 0 1 1 0 0 0 1 0 1 0 0 1

(a) Perform Histogram Equalization for following. Obtain a plot of original as well as Equalized Histogram. [10] 0 1 2 3 4 5 6 7 Intensity 70 100 40 60 0 80 10 40 No.of Pixel (b) Write DDA algorithm to draw a line. Also explain the basic transformation in Computer Graphics. [10] [10] (a) Assume that the edge starts in the first row and ends in the last row for the following gray image Row Column 5 6 7 6 7 1 1 0 3 [10]

3.

Sketch all possible paths and determine the edge corresponding to minimum cost path. (b) Apply KL transform on. Given X1 = (0, 0, 0)T X2 = (1, 0, 0)T X3= (1, 1, 0)T X4 = (1, 0, 1)T 4. Explain in detail following image enhancement tech in spatial domain : (a) Image negative (d) High pass filter (b) Bit plane slicing (e) Low pass filter (c) Contrast stratching (f) Median filter

[20]

5.

(a) Find Huffman code for following six symbols. The symbols and their probabilities are given in tabular form. [10] Symbol Probability a1 0.1 a2 0.4 a3 0.06 a4 0.1 a5 0.04 a6 0.3

Also find it average length and entropy. (b) Consider the image A= 0 0 0 1 1 1 1 0 1 0 0 1 0 0 0 0 0 [10]

Let the structuring element B = 6.

perform (1) Erosion, (2) Dialation, (3) Opening. [10] [10] [20]

(a) Explain Hough transform technique for detection of shape from image edges. (b) Explain Shape No. and find the Shape No. for code 011103022232212111

7.

Write short notes: (a) Wavelet transform (b) Psychovisual redundancy (c) Split and merge (d) Polygonal approximation (e) Euler No

University Question Papers

Image Processing May 2007

Time : 3 Hrs.] N.B.: (1) (2) (3) (4) Question No. 1 is Compulsory. Attempt any four out of remaining six questions. Figures to the right indicate full marks. Assume suitable data if necessary. [Marks : 100

1.

Justify/contradict following statements (any five) : [20] (a) Chain codes can be made invariant to translation and rotation. (b) Median filter is the best solution to remove salt and pepper noise. (c) All image compression techniques are invertible. (d) If Kernel of an image transform is separable and symmetric, the transform is expressed in matrix form. (e) Walsh transform is nothing but sequency ordered Hadamard matrix. (f) Image resulting from poor illumination cannot be segmented easily. (a) A 64 64 image, represented by 8 bits/pixel has following gray level distribution : Gray Level No. of Pixel 0 128 1 75 2 280 3 416 4 635 5 1058 6 820 7 684 [10]

2.

Perform Histogram Equalization and give new distribution of gray level. Show plots of original and equalized image. (b) Explain basic principles of detecting following in the images. [10] (i) Points (ii) Lines (iii) Edges. Give 3 3 mask for each of them and explain their operation. 3. (a) Given below is the table of 8 symbols and their frequency of occurrence : Symbol Frequency S0 0.25 S1 0.15 S2 0.06 S3 0.08 S4 0.21 S5 0.14 S6 0.07 S7 0.04 [10]

Give Huffman code for each eight symbols. (b) Write DDA algorithm to draw a line. Calculate all points using the same algorithm to draw line between points x(10, 10) and y(20, 15) [10] 4. (a) Show that 2D DCT of an image can be computed by row and column passes with 1D DCT algorithm. [10] (b) For 2 2 transform A and the image U 1 2 1 1 1 A= 1 1 , U = 3 4 2 Calculate transformed image V and the basis image. [10] (a) Explain the method of segmentation of images by Region Splitting and merging. [10] (b) Explain the following morphological operation in terms of dilation and erosion operations : Hit or miss, Open, close, Boundary, Skeleton. [10] (a) Assuming that edge starts in the first row and ends in the last row. For following gray level image, sketch all possible paths and determine edge corresponding to minimum cost path [10] 2 2 0 2 7 1 7 5 5 [10] 2 4 2 2 3 6 4 6 0 7 3 1 [10] [10]

5.

6.

(b) For the 3 bit 4 4 size image perform following operations : (i) Negation (ii) Thresholding with T = 4 1 (iii) Intensity level slicing with 2 background with r1 = 2 and r2 = 5 (iv) Bit plane slicing for MSB 5 and LSB plane 3 (v) Clipping with r1 = 2 and r2 = 5 7. (a) Draw and explain 2D Fast Wavelet Transform Filterbanks. (b) Write short notes on any Two :(i) Fourier Descriptors (ii) Chain Code (iii) Image Compression Standard.

Vidyalankar : BE IP

Image Processing November 2007

Time : 3 Hrs.] N.B.: (1) Question No. 1 compulsory. (2) Answer any four questions out of remaining six. (3) Assume suitable data wherever necessary and state them clearly. [Marks : 100

1.

Justify/contradict the following statements. [20] (a) For digital image having salt and pepper noise, median filter is the best filter. (b) Laplacian is better than gradient for delection of edges. (c) For continuous image, histogram can be perfectly equalized but it may not be so far digital image. (d) Run length coding always gives data compression. (e) Segmentation algorithm for monochrome image generally based on 2 basic properties of grey level values. (a) Explain a DDA algorithm. Using DDA algorithm draw a line. Calculate all points between point x(10 ,12) and y (20, 20). [10] (b) Define : (i) Eucleadean distance [2] (ii) City block distance [2] (iii) Chess board distance [2] (iv) 4, 8, m. connectivity [4] (a) What are the different types of redundancies in digital image ? Explain in brief. [8] (b) Explain any four method of data compression. Specify whether it is lossy or lossless and which type of redundancy it will minimize. [12] (i) Adaptive delta Modulation (ii) Differential pulse code modulation (iii) Transform Coding (iv) Run length coding on bit planes (v) LZW algorithm. (a) (b) (c) (d) State and explain properties of 2D Fourier Transform. Write the expression for DFT of 2D signal. Explain Homomorphic filter in detail. Find the DFT of the given image. 0 1 2 1 1 2 3 2 2 3 4 3 1 2 3 2 [10] [5] [5] [8] [4] [4] [4]

2.

3.

4.

5.

(a) Explain filtering in spatial domain. (b) Explain histogram equalization. Derive the transformation function and draw new histogram. (c) Equalize the following histogram. Pr(r)

r 1 6. (a) Given image has the following intensity distributions where P1(z) corresponds to intensity of objects and P2(z) corresponds to the intensity of the background. Assume P1 = P2 and find the optimal threshold between object and background Pixels. P2(z) [5] P2(z) P1(z) = =

1 Z +1 2 1 1 Z 2 2

P1(z)

0 1 2 3

(b) Define Erosion and Dilation operators. Explain their utility in image processing. (c) For the following boundaries shown below. Find the order, Chain code, first difference and shape no.

[5] [5]

(d) Show that the Laplacian mask is Isotropic filter. (Use 3 3 mask).
10

[5]

University Question Papers

7.

Write short notes on (any five) : (a) Discrete Cosine Transform (d) Moments

[20] (b) Sampling and Quantization (e) Hit and Miss Transform (c) Hough Transform (f) Vector Quantization

Image Processing May 2008

Time : 3 Hrs.] N.B.: (1) Question No. 1 is Compulsory. (2) Attempt any four questions out of remaining six questions. (3) Assume suitable data and draw diagram if necessary. [10] [10] [Marks : 100

1.

(a) Write DDA algorithm to draw a line. Also explain the basic transformations in Computer Graphics. (b) Explain following terms: (i) Image Processing and Graphics (ii) Brightness Adaptation (iii) Nonuniform Sampling (iv) Isopreference Curves (v) Fourier Descriptor.

2.

(a) Show that 2DDFT of an image can be computed by row and column passes with 1DDFT algorithm. [10] (b) Explain DCT and its applications. Apply DCT on the given image. [10] 2 1 1 2 0 1 0 1 1 0 0 2 0 1 1 3

3.

(a) What is an edge. Give typical edge profiles and convolution masks for Roberts, Laplace and Prewitt edge detectors. [10] (b) Explain following methods of image segmentation by giving appropriate illustrations: [10] (i) Region growing (ii) Splitting and Merging What is the importance of image segmentation in satellite images and biomedical images. (a) Explain in detail the effects of opening and closing operations. How does the output of opening and closing changes as a function of the shape of structuring element? Consider suitable image for explanation. [10] (b) Explain HoughTransform and its applications in detection of shapes. [10] (a) Explain Filtering in Spatial domain. (b) A 64 64 image represented by 8 bits/pixel has following gray level distribution. Gray level No. of pixels 0 1 2 3 4 5 6 7 686 [10] [10]

4.

5.

120 80 280 420 650 1060 800 Perform Histogram Equalization and show plots of original and equalized image. 6.

(a) What are different types of redundancies in digital images and give methods to remove those redundancies. (b) Explain the following with example and its applications: (i) Chain Code (ii) Moments Write short note on: (a) Wavelet Transform (b) Edge linking and boundary detection via graph theoretic technique (c) Image compression standards (d) Homomorphic filters.

[10] [10]

7.

[20]

11