Self-adaption harris corner detection algorithm based on image contrast area

摘要

Since threshold value of Harris algorithm can only be set by human by virtue of experience and corner point cannot be effectively extracted in different image contrast areas, this text proposes a new self-adaption improved algorithm. Firstly, image pixels will be screened by Sudoku method; then, partial mean square error of the screened pixels will be calculated to determine the contrast ratio of the area where the pixel is located to; finally, the optimal matching corner point will be obtained. Besides, experiments are required to prove effectiveness and feasibility of the algorithm.