Question 1)(a) Derive the expression for the Laplacian of Gaussian in 1D.
(b) Develop from fundamentals, the Laplacian-based edge-detection algorithm.
(c) What is edge-linking, and what is its significance? Describe an edge-linking algorithm.
Question2)(a) Describe an algorithm to detect the presence and location of elliptical contours in a binary image. Write down the pseudo-code for implementing the algorithm on a computer.
(b) An object in a digital image is located with its centroid at (5.6,12.5). Describe clearly, all the step (10) s in implementing the following transformations on the object:
(i) clockwise-rotation about its centroid, by 60 degrees, and
(ii) shrink it by a factor of 2.5. You are allowed to interpolate only once (you may use the nearest neighbor method).
Question 3)(a) Find the dilation, erosion, opening, and closing, of a circular disc by (i) itself, (ii) a bigger circular disc, and (iii) a smaller circular disc.
(b) An ensemble of images possesses predominantly low-frequency-components. Explain the behavior of the Wiener filter designed to restore an image from the ensemble, from its noisy version (noise may be assumed to be white).
(c) Indicate and describe briefly, an application of moments in the context of medical image processing.
Question 4)(a) Prove that the central moments are shift-invariant.
(b)(i) Define a “connected component”.
(ii) Find the result of connected-component labeling applied on the binary image shown in Figure given below, using:
(i) 4-adjecancy, and
(ii) 8-adjacency. List the steps taken in arriving at the result.
Question 5)(a) Explain the concepts of Hue, Saturation and Intensity. Indicate two advantages of the HSI model over the RGB model.
(b) Discuss the construction of the CIE Chromaticity-diagram.
(c) Indicate and discuss all the steps involved in colour-based segmentation for extracting candidate Mycobacterium tuberculosis in images of ZN-stained sputum-smear.
(d) Discuss the necessity of colour-calibration in the context of wound-image processing.