Abstract
Abstract—In this paper, it presents an improved Harris feature extraction algorithm, which is based on Harris algorithm to extract candidate feature points and screen angular points that satisfy certain constraints on the edge of circular neighborhood for the determination of final feature points. And then it makes use of Gaussian-weighted to establish a proximity matrix between internal feature points in two images and uses the singular value decomposition to get their eigenvectors. At last it makes use of eigenvectors to create a relation matrix to reflect the matching degree between feature points for matching. And it shows that extracted feature points in the proposed method in the experiments is robust strongly for rotation, illumination, noise, etc., with good matching effect, low rate of false matches.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have