A Fast Image Matching Method Based on Improved SURF

Abstract In order to solve the problems of low matching accuracy, slow speed and high system overhead in image methods, a rotation binary descriptor construction method based on Speed Up Robust Features (SURF) feature point detection is designed by using different Fast Library for Approximate Nearest Neighbors (FLANN) parameters filtering mechanism screen out wrong matches according types descriptors constructed extraction algorithms. This ensures scale invariant while simplifying representation speeding up calculation initial stage combining characteristics descriptors. Finally, Hamming distance used as improve success rate final matching. The experimental results show that accuracy improved 1.5% time 0.116s, robustness noise ensured.