MENU

Panoramic Image Stitching

603
0

This project involved implementing the algorithms necessary to stitch images together — which, in summary, meant building upon Harris corner detection to then implement adaptive non-maximum suppression (algorithmically creating an even spread of detected corners), extracting feature descriptors around each corner and finding the most similar corners in a to-be-stitched image, then finally computing the best transformation to stitch one image to the next by implementing RANSAC (an algorithm to randomly choose corner points and vote on how good of a transform they produce)

GitHub

CIS 581 Computer Vision & Computational Photography project

11/09/14