Set No.

Code No: RR411103

IV B.Tech I Semester Supplimentary Examinations, November 2008

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 one of the bio-medical applications of pattern recognition system in

detail.
(b) Classify the pattern x=(4,3) into any one of the three classes characterized
by the following decision function. d12 (x) = x1 x2 + 5, d13 (x) = x1 +3,
d23 (x) = x1 + x2
[6+10]
2. Assume that the following pattern classes have normal probability density functions:
w1 : {(0, 0) , (2, 0) , (2, 2) , (0, 2) }andW2 : {(4, 4) , (6, 4) , (6, 6) , (4, 6) }
(a) Assuming P(W1 ) = P(W2 ) = 12 , obtain the equation of the Bayes decision
boundary between these two classes.
(b) Sketch the boundary.

[12+4]

3. (a) Test the following pattern classes for linear separability using the LMSE algorithm. w1 : (-1,-1) , (0,0) , (1,1) and w2 : (-1,1) , (1,-1) .
(b) Explain the gradient descent approach. What is the criterion function used in
this approach.
[8+8]
4. (a) What is the difference between syntactic and mathematical approaches of
pattern recognition.
(b) Explain the different types of grammars with the help of suitable examples.
[8+8]
5. (a) Discuss the various geometrical transformations of the image function.
(b) Suppose the sequence [2 3 2 1] represents a discrete function. Compute Discrete Fourier transform of the function.
[8+8]
6. (a) Write a note on the following:
i. image subtraction
ii. image averaging

[4+4]

(b) Show that a high pass filtered image can be obtained in the frequency domain
as High pass = original - low pass (assume 3x3 filters)
[8]
7. (a) How do you measure information?
(b) Describe the compression of image by eliminating coding redundancy [4+12]
1 of 2

Set No. 1

Code No: RR411103

8. (a) A binary image contains straight lines oriented horizontally, vertically diagonally at 450 and - 450 . Give a set of 3 x 3 masks that can be used to detect
one -pixel-long breaks in these lines. Assume that the gray level of the lines
is 1 and that the gray level of the background is 0.
(b) Prove that the average value of the Laplacian operator
 2 2
 2
r
r
2
h =
exp 2
is zero.
2
4

2 of 2

[8+8]

Set No. 2

Code No: RR411103

IV B.Tech I Semester Supplimentary Examinations, November 2008

IMAGE PROCESSING AND PATTERN RECOGNITION
(Bio-Medical Engineering)
Time: 3 hours
Max Marks: 80
Answer any FIVE Questions
All Questions carry equal marks

1. Explain in detail about the fundamental problems in the design of pattern recognition system. Also, discuss the methods to overcome them.
[8+8]
2. With a brief note on clustering concepts, explain the maximin distance algorithm
in detail.
[8+8]
3. (a) State and prove perceptron convergence theorem.
(b) Draw the basic perceptron model and explain the reward - punishment concept.
[8+8]
4. Discuss Robbins-monro algorithm for pattern classification and extend the algorithm to multi - dimensional case.
[16]
5. (a) Explain briefly different types image acquisition systems
(b) Give transformation matrices for the following
i. Translation
ii. Scaling
iii. Rotation along z-coordinate.

[8+8]

6. (a) Name and explain different methods available for image enhancement in spatial domain and frequency domain.
(b) Explain the concept of enhancement by point processing using suitable example.
[8+8]
7. Briefly write about the lossless predictive coding and lossy predictive coding. Compare their applications.
[10+6]
8. (a) What is meant by image segmentation? Mention the applications of image
segmentation.
[4+4]
(b) Explain about detection of discontinuities.

1 of 1

[8]

Set No. 3

Code No: RR411103

IV B.Tech I Semester Supplimentary Examinations, November 2008

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 briefly about the character recognition system?

(b) Explain briefly about the speech recognition system?

[8+8]

2. Classify the following samples using the k-means algorithm in to 3 clusters.

{X1 (0, 0), X2 (1, 0), X3 (0, 1), X4 (1, 1), X5 (6, 6), X7 (7, 6), X8 (8, 8), X9 (8, 9)andX10 (9, 9)}.
Explain the necessary steps used in above problem.
[16]
3. (a) Describe the basic perceptron model for separating the patterns of two classes.
(b) Explain how the basic perceptron model can be extended for the multi-class
case.
[8+8]
4. (a) Can you construct context free grammar for even palinrome? If so, write the
grammar? If not give the reasons.
(b) Can you construct a regular grammar for odd palinrome? If so write the
grammar? If not give the reasons.
[8+8]
5. (a) Explain the periodicity, conjugate and rotation properties of Fourier transform.
(b) A real function f(x) can be decomposed as the sum of even and odd functions.
Show that feven (x) = (1/2)[f (x) + f (-x)] and fodd (x) = (1/2)[f (x) - f (-x)]
[8+8]
6. (a) Write a note on the following:
i. image subtraction
ii. image averaging

[4+4]

(b) Show that a high pass filtered image can be obtained in the frequency domain
as High pass = original - low pass (assume 3x3 filters)
[8]
7. (a) How do you measure information?
(b) Describe the compression of image by eliminating coding redundancy [4+12]
8. (a) Discuss the methods for detecting the discontinuities in an image.
(b) Describe the role of various masks used for convolution with an image for edge
detection.
[8+8]

1 of 1

Set No. 4

Code No: RR411103

IV B.Tech I Semester Supplimentary Examinations, November 2008

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) Mention the conditions of case1, case2, and case3 of linear decision functions
to classify the pattern?
(b) Classify the pattern X = (6,5) into any one of the three classes characterized
by the following decision function.
d1 (x) = x1 + x2 , d2 (x) = x1 + x2 -5, d3 (x) = x2 + 1.
[8+8]
2. (a) What is the difference between 1-NN and q-NN rules in pattern classification.
(b) What is the basic principle involved in stylized character recognition system?
[8+8]
3. (a) Explain the abstract steps followed in gradient descent algorithm.
(b) Illustrate geometrically the progress obtained in successive iterations using
gradient descent algorithm.
[8+8]
4. (a) Consider the grammar G = {VN, VT, P, S} where VN = P{S} ,VT = {a, b}
and P = {S aSb, S ab} .Generate the sentence by applying the first
production m-1 times and second production once.
(b) Explain how a statistical recognition problem is formed.

[8+8]

5. (a) Define Discrete Fourier transform and inverse Discrete Fourier transform and
mention some properties of Discrete Fourier transform.
(b) What are the properties of the distance function? Define Euclidean distance,
city block distance and chessboard distance.
[8+8]
6. (a) Explain why the discrete histogram equalization technique will not, in general,
yield a flat histogram
(b) Show that a high pass filtered image in the frequency domain can be obtained
by using the method of subtracting a low pass filtered image from the original
image.
[8+8]
7. (a) How do you measure information?
(b) Describe the compression of image by eliminating coding redundancy [4+12]
8. (a) A binary image contains straight lines oriented horizontally, vertically diagonally at 450 and - 450 . Give a set of 3 x 3 masks that can be used to detect
one -pixel-long breaks in these lines. Assume that the gray level of the lines
is 1 and that the gray level of the background is 0.
1 of 2

Set No. 4

Code No: RR411103

(b) Prove that the average value of the Laplacian operator

 2
 2 2
r
exp 2
is zero.
2 h = r
4
2

2 of 2

[8+8]