Research on location algorithm and idendtify matching of object based on SURF-BRISK(PDF)
长安大学学报（自然科学版）[ISSN:1006-6977/CN:61-1281/TN]
- Issue:
- 2016年03期
- Page:
- 93-101
- Research Field:
- 汽车与机械工程
- Publishing date:
Info
- Title:
- Research on location algorithm and idendtify matching of object based on SURF-BRISK
- Author(s):
- HUI Ji-zhuang; LUO Li; YANG Yong-kui; LIU Qiong
- (Key Laboratory for Highway Construction Technology and Equipment of Ministry of Education, Chang’an University, Xi’an 710064, Shaanxi, China)
- Keywords:
- mechanical engineering; feature extraction; feature matching; MSAC algorithm; affine transformation; location
- PACS:
- TH2
- DOI:
- -
- Abstract:
- To solve the problem that SURF and BRISK algorithm can’t accurately identify work-pieces in the image detection matching, this paper proposed a matching positioning method of SURF-BRISK algorithm combining with hamming distance and affine transformation. The SURF-BRISK algorithm was used to make the first match of image feature extraction to work-piece. The hamming distance was utilized to conduct similarity measure of feature matching after the first match. The MSAC algorithm was applied to eliminate false matching points by the second match. Then, the affine transformation parameters were calculated by affine model. According to the object template centroid and the affine transform parameters, the object centroid coordinates were obtained. By means of the technology of binocular stereo vision calibration and 3D reconstruction principle, the three-dimensional coordinates of the work-piece were gained. It provided the information for the robot accurate grasping. In the experiment, the matching recognition and the 3D location procedures were performed by MATLAB. The messages of the work-piece received by MATLAB and VC++ and OpenCV mixed programming were transmitted to the robot controller. The results show that this method has strong adaptive ability. It can detect the object centroid in the complex environment. It can obtain the target centroid coordinates and implement the real-time accurate grab of the robots to work-piece when ignoring the geometric distortion.
Last Update: 2016-06-12