The objective of this homework is to reinforce the scale space response extrema concept in key points detection, and ask you to compute a LoG pyramid for scale space extrema detection, and compare that with the DoG pyramid based SIFT detection. An example is illustrated below, on the left is an image and its SIFT detection, with vl sift() and peakStrenthThres=16, on the right is the LoG based detection: 50 100 150 200 250 300 350 400 450 FAMILJE 100 200 300 400 500 600 700 50 100 150 200 250 300 350 400 450 100 200 log max 300 400 500 The LoG pyramid is computed for o = {1.200 1.518 1.920 2.429 3.072 3.886 4.915} 600 700 100 200 300 400 100 200 300 400 LOG (1.2) 200 V P 400 600 LOG (2.4) KANT WIREL 200 400 600 100 200 300 400 100 200 300 400 LOG (1.5) PAMARI TONG HAR 200 400 600 LOG (3.1) ENCE 200 400 600 100 200 300 400 100 200 300 400 LOG (1.9) UND LA 200 CHEM 200 LOG (3.9) LENO 400 600 Mpy DORE 400 600 You may choose to do SIFT like spatial-scale neighborhood extrema detection, or you may use the ALP like polynomial fitting and direct extrema detection method. The ALP like method is illustrated in the figure below: 15 10 5 0 1 1.5 LoG response 2 0 2.5 3 3.5 Q1 [20 pts] Compute LoG Pyramid, show your code function [log_pyramid]=getLoGPyramid (im, scales) wwwwwwwwwwww 20 0 -20 -40 -60 -80 -100 -120 -140 -160 1 1.5 LoG response diff 2 0 2.5 3 The LoG response is given asf(o)=ao + bo + co +d, the fitting can be done with polyfit(x, y, 3), and you can also compute its gradient. An extrema is detected by checking its gradient value. 3.5 Q2 [30 pts] Compute LoG response extrema. ECE/CS 479/5582 Computer Vision function [x0, y_max]=getScale Response Extrema(x, y) HW-3 Q3 [50 pts] Give the image level LoG key points detection and visualize and compare with SIFT detection. function [x, y, scale] =getLoGKeyPoints (im, log_peak_thres)