Characteristic Point Match Algorithm Based on the SURF in Binocular Stereo Vision

Abstract

One of the main components of the computer vision is 3D reconstruction. Extracts and the match based on two-dimensional picture characteristic point is the 3D reconstruction technology core. Taking binocular stereoscopic vision theory as a foundation, through extracts the characteristics of SURF based on the multi-scale analysis, this feature has specific scale reproducibility, put the epipolar constraints and disparity constraints as conditions judgement to specific screening, it is greatly reduced the search range. The experiment use the matching method for three different actual scenes, the results shows that the method can enhance the image matching speed and precision, and can get more precise dense parallax, used for reconstruction complete scene.