Image Processing May 2005

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

1.

Reason the following contentions [20] (a) Enhancement process does not add any information to the image. (b) Storing of images in frame buffers has many advantages over storing them in computer memory. (c) The process of removing the psychovisual redundancy is a lossy process. (d) If the Kernel of an image transform is seperable and symmetric, the transform can be expressed in matrix form. (a) What are Fidelity Criteria ? How can they be used in quantifying the nature and extent of information loss ? (b) Briefly describe the salient features of the following (i) Fourier descriptors (ii) Run length coding (iii) WalshHadamard transform. State and explain the difference between the following (a) Uniform and Non uniform sampling (b) Convolution and correlation (c) Spatial domain processing and Transform domain processing (d) Lossless and lossy compression. (a) Give the 3 3 marks for the following (i) Prewitt (ii) Sobel. (b) Write short note on : (i) Contrast stretching (ii) Histogram Specification (iii)Homomorphic filtering. [5] [15]

2.

3.

[20]

4.

[2] [18]

5.

(a) Explain Isoperference curves with respect to digital image processing. [5] (b) What is image segmentation ? Explain the following methods of image segmentation giving appropriate explains. [15] (i) Region growing (ii) Splitting and Merging (iii) Thresholding. Explain (a) Hit or Miss Transform. (c) Moments. [20] (b) Skeletonization using Medial Axis Transform. (d) Chain code.

6.

7.

(a) Plot the histogram for the following image. Equalize the histogram and then plot the equalized histogram and the image corresponding to the equalized histogram. The 4 5 image is represented by 3bpp. [10] 7 3 4 2 1 3 2 4 2 5 4 6 2 5 0 2 2 7 0 3 [10] (ii) Texture.

(b) Explain (i) Arithmetic Coding

Image Processing November 2005

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

1.

Reason the following contentions [20] (a) The first difference of chain code normalise to Rotation. (b) Spectral approach using Fourier spectrum is useful in texture matching. (c) Brightness discrimination is poor at low levels of illumination. (d) Quality of picture depends on the number of pixels and number of grey level that represent the picture. (a) A contrast stretching is performed as per the following characterize =7 3m5 =u otherwise. where u and v are the intensities for the input and output image respectively. Determine the o/p image when, contrasts stretching g is applied on following 0 1 0 2 1 6 2 3 5 5 1 6 2 4 4 3 2 1 1 3 5 3 0 1 2 1 2 6 6 2
1

2.

[12]

Vidyalankar : B.E. IP

(b) State and explain the features of median filtering. Complete the o/p of median filtering in the following case. [8] (i) y (m) = {2 4 8 3 2} and w = {1 0 1 2} (ii) y (m) = {8 2 4 3 4} and w = {1 0 1} 3. (a) Three column vectors are given below w0 w0 w0 0 = w 0 1 = w1 2 = w 2 0 2 4 w w w where w = ej2/3 Prove that they are orthogonal. (b) Using these vector generate 3 3 nine orthogonal patterns pij (x, y), I, j, x, y = 0, 1, 2. (c) Evaluate the transform coefficient aij so that given image I(x, y) is given by
I ( x, y ) = a ij Pij ( x, y )
i = 0 j= 0 2 2

[6]

[6] [8]

where I(x, y) =

3 1 2

2 3 1

1 2 3

4.

(a) What is histogram of a digital image ? What information one can get by observing the histogram of image ? [8] (b) Histogram of a digital image with eight quantization level is shown below perform histogram equalization. [12] Derive the transformation function and new histogram. Grey level r 0 1 2 3 4 5 6 7 No. of grey level hr 220 140 50 60 70 170 130 160 (a) Give following masks of size 3 3 and explain their usefulness in image processing [12] (i) Sobel operator (ii) Roberts operators (iii) Prewitt operator (iv) High pass filter (v) Median filter (vi) Freichen operator. (b) An image of size 5 5 quantized to 16 grey levels is shown below. Operate on this image by any two edge extraction operators of size 3 3. Use appropriate thereshold should value. Give output before and after theresholding. Give your comments. [8] 10 12 15 15 2 12 10 15 3 8 15 14 2 10 10 13 4 15 12 2 1 12 14 1 1 (a) What are different types of redundancies in digital images ? (b) Explain following method of data compression indicating clearly whether they are lossy or lossless. Also state the type of redundancy they can minimize (any four) (i) Huffman coding (ii) Transform coding. (iii) Adaptive delta modulation (ADM). (iv) Differential pulse code Modulation (DPCM). (v) Runlength Coding on bit planes. Write short notes on (five any) (a) JPEG. (c) Discrete Cosine transform. (e) Moments. (b) Fourier Descriptor . (d) Hit and Miss transform (g) Shape number. [8] [12]

5.

6.

7.

[20]

Image Processing May 2006

Time : 3 Hrs.] N.B. (1) Question No. 1 is compulsory. (2) Attempt any four questions from remaining six questions. (3) Assume suitable data wherever necessary and state it. [6] [8] [6] [20] [Marks : 100

1.

(a) What do you mean by sampling? What are the different types of sampling? (b) Explain the terms Brightness adaptation and discrimination with reference to human visual system. (c) State and prove symmetry and periodically property of DFT. Justify / Contradict the following statements (a) For digital images having salt pepper noise, median filter is the best filter. (b) The first difference of a chain code normalize it to rotation. (c) Minimum cost algorithm is meant for detection of edges. (d) Laplacian is better than gradient for detection of edges.

2.

University Question Papers

3.

(a) Describe the general compression system model. (b) Explain briefly the Isopreference curves with respect to Digital Image Processing. (c) Define Moments. Normalized moments and central moments. Write short notes on the following : (a) Redundancies in digital images (c) Homomorphic Filter

[8] [6] [6] [20]

4.

(b) Fidelity criteria (d) Texture analysis

5.

(a) Histogram of a digital image with 8 quantization levels is given below. Perform Histogram Equalization. Derive the transformation function and the new histogram and draw them. [10] 0 1 2 3 4 5 6 7 (Gray level ) r 175 12 250 98 153 35 220 57 (N0. of pixels with gray level r) nr (b) Explain with suitable example the LZW coding and decoding technique for image Compression. [10] (a) What is segmentation? Explain the method of segmentation by region splitting and merging. (b) Poorly illuminated images are the difficult ones to be segmented. Explain. (c) Can variable length coding procedures be used to compress a histogram equalized image with 2n levels. Explain. Write short notes on: (a) Delta Modulation and differential pulse code modulation (DPCM). (b) Smoothing and sharpening Filters (c) Gamma Correction (d) RGB color model [8] [4] gray [8] [20]

6.

7.

Image Processing November 2006

Time : 3 Hrs.] N.B. (1) Question one is compulsory. (2) Attempt any 4 questions from remaining six questions. (3) Assume suitable data wherever necessary and state it. [20] [Marks : 100

1.

Write short notes on (a) Applications of Image Processing. (b) Image formation in the eye. (c) Correspondence between Filtering in the Spatial and Frequency Domains (d) Hough Transform. (a) Write a note on Wavelet transform (b) Classify Image Compression Techniques (c) Name and explain any four zero memory operations (a) What do you mean by Image Averaging? (b) Texturean Image descriptor. Explain. (c) Can Graph theoretic techniques be used in Image Segmentation? If so, how? (a) What do you understand by Variable Length Coding techniques? Explain at least two techniques. (b) What is constant area coding technique? (c) What is Chain code and what are its applications.

2.

[5] [5] [10] [5] [10] [5] [10] [5] [5]

3.

4.

5.

(a) Define and explain Moments, Normalized moments, and central moments. [10] (b) What are Morphological operations? For a region explain Boundary extraction operation and then region filling operation on the extracted boundary. [10] (a) How do you get a Skeleton of a region using Medial Axis Transformation. [10] (b) Histogram of a digital image with 8 quantization levels is given below. Perform Histogram Equalization. Derive the transformation function and the new histogram. [10] (Gray Level)r (No. of pixels with gray level r) nr 0 250 1 550 2 450 3 250 4 100 5 150 6 150 7 50 8 25 9 25

6.

7.

(a) Write a note on Discrete Cosine Transform and its application in Image Compression. [10] (b) Write 8 8 Haar transform matrix and its Signal Flow Graph. Using this Butterfly diagram (Signal Flow Graph), Compute Haar transform for x(n) = {1, 2, 3, 4, 1, 2, 3, 4}. [10]

Vidyalankar : B.E. IP

Image Processing December 2007

Time : 3 Hrs.] [Marks : 100 N.B. : (1) Question No. 1 compulsory. (2) Attempt any four out of remaining six questions. (3) Figures to right indicate full marks. (4) Assume suitable data if necessary. 1. (a) Define (i) Spatial Resolution (ii) Intensity Resolution (iii) PSF (Point Spread Function) (iv) Dilation (v) Image Restoration (b) (i) Brightness discrimination is poor at low levels of illuminations. (ii) The first difference of Chain Code normalise it to rotation. (iii) Enhancement process do not change the information content of image. (iv) Quality of picture depends on the number of pixels and arey level that represents the picture. (v) For digital image having salt pepper noise, median filter is the best filter. 2.

[10]

[10]

(a) Write a note on Discrete Cosine Transform and its application in Image processing. [10] (b) Write 8 8 Haar transform matrix and its signal flow graph. Using this Butterfly diagram. Compute Haar transform for x(n) = {1,2,3,4,1,2,3,4}. [10] (a) Define and explain Moments, Noramlised moments and Central moments. [10] (b) What are Morphological operations ? For a region, explain Boundary extraction operation and then region filling operation on the extracted boundary. [10] (a) What is impulse response of each filter ? (i) Transfer function is (ii) Frequency response is : H(w1 , w 2 ) = 1 2 cos w1 2 cos w 2 (b) Discuss Enhancement Technique in spatial domain used for images.
H(Z1 , Z2 ) = 1 a1z1 1 a 3 z 2 1

3.

4.

[10]
a 4 z1.z 1 2

[10]

5.

(a) Write an expression for a 2D DFT. What is its relationship with one dimension DFT ? How onedimensional FFT algorithms can be used to compute two dimensional DFT of an digital image. [6] (b) Using 4 point FFT algorithms. Calculate 2D, DFT of. [8] 0 0 3 1 1 1 2 2 f(x, y) = 2 2 1 3 1 1 2 4 (c) Illustrate opening and closing operations with suitable example. [6] (a) What are different types of redundancies that can be present in a digital image and state which method can be used to remove/reduce them. [8] (b) What is Morphology ? Name and explain the basic operations in morphology. [8] (c) Write a short note on Wavelet transform. [4] Write a short notes on (any five) : (a) Texture (b) Arithmetic Coding (d) Fourier Descriptors (e) RGB Colour model (g) Uniform and Nonuniform samling. [20] (c) Hit or Miss Transform (f) Run length coding

6.

7.

Image Processing May 2008

Time : 3 Hrs.] N.B. (1) (2) (3) (4) Question No. 1 is Compulsory. Attempt any four from the remaining six questions. Figures to the right indicate full marks. Assume suitable data if necessary. [20] [Marks : 100

1.

Justify/contradict the following statements: (a) Lossy compression of the image makes the image loose smooth information. (b) Shape number uniquely describes an object. (c) The zeroth (LSB) plane of the bit plane slicing contains high frequency information of the image. (d) Image subtraction is used for scene matching and detection. (e) Convolution in spatial domain is multiplication in Fourier domain. (a) Explain what is ringing effect. Explain how butterworth filters remove the ringing effect. (b) Explain the following geometrical measurement features for shape representation: (i) Perimeter (ii) Holes (iii) Euler number (iv) Eccentricity.
4

2.

[8] [12]

University Question Papers

3.

(a) Explain the behaviour of the first order derivative and second order derivative functions at the point, ramp, smooth and step changes of the gray values. [8] (b) Explain and derive the successive doubling method used in FFT. Find and compare its computational complexity with DFT. [12] (a) Explain the use of finding connectivity of pixels. Explain the methods to find the connectivity of pixels and the distance between the two pixels. [8] (b) Explain Hough transform. Using it, detect the line for the following Image: [12] 1 0 0 0 0 0 0 1 0 1 0 0 1 0 0 1 0 0 1 0 0 0 1 0 1 (a) Explain the different operators used for segmentation of an image. (b) Define and perform opening, closing, boundary detection operation on the following image using as the structuring element: 1 1 1 1 1 1 1 1 1 1 1 1 1 [8] [12]

4.

5.

6.

(a) What is the skeleton of an image? Explain the methods to find the skeleton of an image. [8] (b) What are moments? Why are they used in image processing? Explain Using moments find the centroid of the following image: [12] 0 0 0 0 0 0 1 1 1 0 0 1 1 1 0 0 1 1 1 0 0 0 0 0 0 (a) Explain DCT and its properties. Find the DCT for the following image: 2 4 4 2 4 6 8 3 2 8 10 4 3 8 6 2 (b) Explain JPEG and compress the above image using JPEG. [10]

7.

[10]

Image Processing November 2008

Time : 3 Hrs.] N.B. (1) Question No. 1 is compulsory. (2) Attempt any four out of remaining six questions. (3) Assume suitable data wherever necessary and state it. [20] [Marks : 100

1.

Justify/contradict following statements : (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) Image subtraction is used for scene matching and detection. (e) Image resulting from poor illumination cannot be segmented easily. (a) A 64 * 64 image, represented by 8 bits/pixel has following gray level distribution :

2.

[10]

0 1 2 3 4 5 6 7 Gray Level 128 75 280 416 635 1058 820 684 No. of Pixel 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) 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]
5

Vidyalankar : B.E. IP

4.

(a) What are different types of redundancies in digital images and give methods to remove those redundancies? (b) How many calculations do we require to compute the Hadamard transform? If X(n) = {4, 2, 2, 4} (c) State and prove symmetry and periodically property of DFT. (a) Given a 4 4 image whose gray level, orders lexicographically are as follows: 3 1 2 0 2 0 3 1 2 3 1 0 1 3 0 2 calculate the spatial moments up to second order. (b) What do you understand by image averaging? (c) What is chain code and what are its application?

[10] [4] [6] [10] [5] [5]

5.

6.

(a) What do you understand by Discrete Cosine Transform? Explain its application in image compression. [10] (b) Explain texture and image descriptor. [10] Write short notes on: (i) Wavelet transform (ii) Edge linking and boundary detection via graph theoretic technique (iii) Image compression standards (iv) Homomorphic filtering [20]

7.