No. of Pages: 2 B.
APJ ABDUL KALAM TECHNOLOGICAL UNIVERSITY
THIRD SEMESTER M.TECH DEGREE EXAMINATION, DECEMBER 201 7

Electronics and Communication Engineering

Signal Processing

01 EC 73 1 5: Computer Vision

Answer any twofull questions from each part Limit
answers to the required points.

Max. Marks: 60 Duration: 3 hours
PART A

a. Explain in detail the steps for computing scale-invariant feature transform.

b. Prove that difference of Gaussian (DOG) function provides a close approximation to the
scalenormalized Laplacian of Gaussian (LOG). 2

2. a. Explain the k- means and mixture of Gaussians approaches for image segmentation. Highlight
5 the difference between the two.

b. What is Laplacian of an image and derive an appropriate kernel for the same. Explain how
4 Laplacian Can be used for edge detection.

3: a. Explain the use Of Hough transform for line detection. 5
b. Write notes on camera intrinsic and extrinsic parameters. 4
PART 8
4. a.What is epipolar constraint? 2

b. Detail the steps involved in projective reconstruction of 3D structure from photos taken by
7 unknown camera.

5. a. What is aperture problem in optical flow?