Research on Identify Matching of Object and Location Algorithm Based on Binocular Vision

This paper proposes a matching positioning method of SURF-BRISK algorithm combining with hamming distance and affine transformation. Using SURF-BRISK algorithm makes the first match of image feature extraction, and hamming distance conducts similarity measure of feature matching after first match, and MSAC algorithm makes the second match to eliminate false matching points. The affine transformation parameters are calculated with affine model. Finally, the object centroid coordinates in complex environment are obtained according to the object template centroid and affine transform parameters, and three-dimensional coordinates of the work-piece are got through the technology of binocular stereo vision calibration and 3-D reconstruction principle, which provide information for robot accurate grasping. Results demonstrate that this method has strong adaptive ability and can detect the object centroid in complex environment, obtain the target centroid coordinates and implement real-time accurate grab of robots to work-piece.