Computer Vision I Homework 2

(Oct 2016 - Nov 2016)

Oct 2016 – Nov 2016

For the written part of this homework, I need to do some calculations based on Epipolar Geometry. On the other hand, for the coding part, there are mainly three tasks: draw bound boxes, edge detection, and sparse stereo matching. In the first task, I need to figure out the proper bounding box on given images. To do so, I need to plot intensity graph and regard the local maximum intensity paint as the center of the bounding box. We also have pretty good accuracy on realistic images. In the second task, a sequence of operations are needed: smoothing, gradient computation, non-maximum suppression, and hysteresis thresholding. As to the last task, corner detection and Sum of Squared Differences (SSD) matching algorithm are used. To further improve the matching result, Epipolar Geometry based matching is needed. Codes are also attached in the report.

Leave a Reply

Your email address will not be published. Required fields are marked *