Set No. 1
Code No: RR4111 RR411103 03
IV B.Tec B.Tech h I Semest Semester er Regula Regularr Examin Examinati ations ons,, Nov Novem ember ber 2007 2007 IMAGE PROCESSING AND PATTERN RECOGNITION (Bio-Medical Engineering) Time: 3 hours Max Marks: 80 Answer any FIVE Questions All Questions carry equal marks ⋆ ⋆⋆⋆ ⋆
1. Explai Explain n various arious design design concepts concepts of pattern pattern recognit recognition ion system. system. Discus Discusss various arious metho ds of implementation. [8+8] 2. Consider an M-class, one dimensional pattern recognition problem in which each class is characterized by a Rayleigh probability density function: p(X /wi ) =
x
√
σi
2π
exp −x2 2σi2
0,
,x≥0
x<0
Deri Deriv ve the Bayes Bayes decis decisio ion n functi function, on, assu assumi ming ng zero zero and and one one loss loss functi function. on. Let 1 p(wi )= m . [16] 3. Find the solution vector ω using reward punishment algorithm to separate the two classes, where the patterns in each class are given by ω1 : {(0, 0, 1) , (0, 1, 1)1 } and ω2 : {(1, 0, 1)1 , (1, 1, 1)1 } [16] 4. (a) Explai Explain n the least-me least-mean an squaresquare-erro errorr algorithm algorithm with with an example. example. (b) Explain the convergence proof oof of the LMSE algorithm.
[8+8]
5. (a) Discuss Discuss the vario various us geometric geometrical al transform transformatio ations ns of the image function function.. (b) Suppose Suppose the sequenc sequencee [2 3 2 1] represen represents ts a discret discretee function function.. Compute Compute Discrete Fourier transform of the function. [8+8] 6. Explain in detail the different derivative operators used for image sharpening.[16] 7. (a) How How do you you measur measuree inform informati ation? on? (b) Describe Describe the compression of image image by eliminat eliminating ing coding redundancy redundancy [4+12] 8. (a) Desc Describ ribee the signi significa ficance nce of Lapl Laplaci acian an operator operator.. Expl Explai ain n with with an examp example le,, how the edges are detected with this operator. (b) (b) Com Compa pare re the the perfo perform rman ance ce of vario arious us edge edge dete detect ctin ingg opera operato tors rs.. ⋆ ⋆⋆⋆ ⋆
1 of 1
[8+8 [8+8]]
Set No. 2
Code No: RR4111 RR411103 03
IV B.Tec B.Tech h I Semest Semester er Regula Regularr Examin Examinati ations ons,, Nov Novem ember ber 2007 2007 IMAGE PROCESSING AND PATTERN RECOGNITION (Bio-Medical Engineering) Time: 3 hours Max Marks: 80 Answer any FIVE Questions All Questions carry equal marks ⋆⋆⋆⋆⋆
1. (a) Explain Explain the mathematical mathematical model model of simple simple pattern pattern recognition recognition system indetail? indetail? (b) (b) How distan ancce fun uncction tionss are use useful in patte attern rn recogn cogniition?
[8+8 8+8]
2. (a) Write rite a short short note on on cluster cluster analy analysis sis.. (b) Explain a cluster-seeking algorithm in detail.
[8+8]
3. Explain Explain the proof of convergence convergence applicable applicable to reward-puni reward-punishmen shmentt algorithm. [16] 4. Discuss Robbins-monro algorithm for pattern classification and extend the algorithm to multi - dimensional case. [16] 5. Write short notes on: (a) Pixel neighbors. neighbors. (b) Pixel connectivity connectivity.. (c) Distance Distance measure. measure. (d) Equivalence of pixels.
[4+4+4+4]
6. (a) Write rite a note note on the follo followin wing: g: i. image subtraction ii. image averaging
[4+4]
(b) Show that a high pass filtered image image can be obtained obtained in the frequency domain as High pass = original - low pass (assume 3x3 filters) [8] 7. (a) Briefly Briefly expl explain ain about about image image compress compression ion.. (b) How image compression is helpful in medical imaging?
[8+8]
8. (a) A binary image image contains contains straight straight lines oriented oriented horizontally horizontally,, vertically vertically diagodiago0 0 nally at 45 and - 45 . Give a set of 3 x 3 masks that can be used to detect one -pixel-l -pixel-long ong breaks in these these lines. lines. Assume Assume that the gray level level of the lines is 1 and that the gray level of the background is 0. (b) Prove Prove that the average average value value of the Laplacian operator 2
∇ h
=
r
2
−σ 2
σ
4
exp
r
2
− 2σ2
is zero.
⋆⋆⋆⋆⋆
1 of 1
[8+8]
Set No. 3
Code No: RR4111 RR411103 03
IV B.Tec B.Tech h I Semest Semester er Regula Regularr Examin Examinati ations ons,, Nov Novem ember ber 2007 2007 IMAGE PROCESSING AND PATTERN RECOGNITION (Bio-Medical Engineering) Time: 3 hours Max Marks: 80 Answer any FIVE Questions All Questions carry equal marks ⋆⋆⋆⋆⋆
1. Explai Explain n various arious design design concepts concepts of pattern pattern recognit recognition ion system. system. Discus Discusss various arious metho ds of implementation. [8+8] 2. (a) What is the relatio relation n between between error error probabili probability ty of 1-NN 1-NN rule and Baye Bayess probability of error. (b) Explain Explain the stylized character character recognition recognition system with the help of flow chart. [8+8] 3. Apply the perceptron algorithm algorithm to find the solution vector vector for the following pattern pattern classes: 1 1 1 1 Sketch h the decisio decision n surface surface obtained, obtained, ω1 : {(0, 0) , (1, 0) }andω2 : {(0, 1) , (1, 1) }. Sketc and indicate the po possitive side of the separating surface. [10+6] 4. (a) Explain Explain the incremen incrementt correction correction algorithm algorithm for for trainable trainable pattern pattern classifier classifier using stochastic approach and derive the conditions for decision function. (b) Explain Explain LMSE algorithm algorithm using stochastic approach approach for pattern classification. classification. [8+8] 5. (a) Explai Explain n the periodicit periodicity y, conjugat conjugatee and rotation rotation properties properties of Fourier ourier transform. (b) A real function function f(x) can be decomposed as the sum of even even and odd functions. Show that f even even (x) = (1/2)[f (x) + f (-x)] and f odd odd (x) = (1/2)[f (x) - f (-x)] [8+8] 6. (a) Explai Explain n the meaning meaning of histogr histogram am equaliz equalizati ation. on. Give Give an example example where where the histogram equalization technique compresses the discrete gray level range of an image. (b) Show that crispened image can be obtained by convolvin convolvingg it with the mask 0 -1 0 -1 5 -1 [8+8] 0 -1 0 7. (a) Draw Draw the image image compress compression ion model model and describe describe the functi functions ons of each each unit in it. (b) What are the various coding techniques techniques for error-free error-free compression and discuss discuss in detail Huff-man co ding. [8+8] 8. (a) What is meant meant by by image image segment segmentatio ation? n? Mention Mention the the applicati applications ons of image image segmentation. [4+4] 1 of 2
Set No. 3
Code No: RR4111 RR411103 03 (b) Explain ab out detection of discontinuities. ⋆⋆⋆⋆⋆
2 of 2
[8]
Set No. 4
Code No: RR4111 RR411103 03
IV B.Tec B.Tech h I Semest Semester er Regula Regularr Examin Examinati ations ons,, Nov Novem ember ber 2007 2007 IMAGE PROCESSING AND PATTERN RECOGNITION (Bio-Medical Engineering) Time: 3 hours Max Marks: 80 Answer any FIVE Questions All Questions carry equal marks ⋆⋆⋆⋆⋆
1. (a) Expl Explai ain n in detail detail about pattern patternss and patter pattern n class classes es with exampl examples es.. Also Also,, explain how the pattern classes are useful for recognition. (b) With a neat block diagram, explain explain a pattern recognition recognition system in detail. [8+8] 2. (a) Write rite a short short note on on cluster cluster analy analysis sis.. (b) Explain a cluster-seeking algorithm in detail.
[8+8]
3. Find the solution vector ω using reward punishment algorithm to separate the two classes, where the patterns in each class are given by ω1 : {(0, 0, 1) , (0, 1, 1)1 } and 1 1 [16] ω2 : {(1, 0, 1) , (1, 1, 1) } 4. (a) Can you you construc constructt context context free free grammar grammar for even even palinrom palinrome? e? If so, write write the grammar? If not give the reasons. (b) Can Can you you constr construct uct a regul regular ar gramm grammar ar for for odd pali palinr nrome ome?? If so write write the grammar? If not give the reasons. [8+8] 5. (a) Give Give the transformation transformation matrix used to to rotate rotate an image by 450 clockwise. How would this transformation is used to achieve the desired image rotation (b) (b) Wr Write short ort no note tess on sampling & quan anttizati ation of an image. 6. (a) Explain in detail image averaging.
[8+8] [4]
(b) What is the use of smoothing smoothing the image? image? What are the differen differentt methods to smo othen the image? [5+7] 7. (a) How How do you you measur measuree inform informati ation? on? (b) Describe Describe the compression of image image by eliminat eliminating ing coding redundancy redundancy [4+12] 8. (a) What is meant meant by by image image segment segmentatio ation? n? Mention Mention the the applicati applications ons of image image segmentation. [4+4] (b) Explain ab out detection of discontinuities. ⋆⋆⋆⋆⋆
1 of 1
[8]
