Research On Industrial Robot Target Recognition And Location Based On Binocular Stereo Vision

With the development of machine vision technology, binocular stereo vision has been widely applied to the workpiece recognition, robot navigation, industrial detection and vehicle recognition, etc.Relative monocular vision, binocular stereo vision technology is available in many cases the flexibility to obtain three-dimensional information of the target,has very obvious advantages, its key steps are: camera calibration, stereo matching and 3 d reconstruction.Based on binocular stereo vision technology, the target workpiece identification and robotic grasping is studied, mainly involving the camera calibration,identification of target artifact matching and robot to the target artifact of grab, etc.First, the camera calibration is realized.Without considering the distortion of the camera case, the method based on linear perspective model calibration, and through the experiment the method has higher calibration accuracy, the accuracy can meet the requirements of workpiece grabbing.Secondly, based on the SURF- BRISK algorithm was presented to match artifacts.Image feature points, were detected by SURF algorithm, BRISK calculated descriptors, and complete the initial match of template images and the target picture.Then the MSAC algorithm was used to finish the second fine match to improve the matching accuracy.Finally, by 3 d reconstruction obtain target centroid of three-dimensional coordinates,workpiece precision robots grasping for parts is realized.Using the template centroid coordinates and affine transformation model to obtain the centroid coordinates of target artifact and combining with binocular stereo vision calibration technology for 3 d reconstruction of target artifact, three-dimensional coordinates of the workpiece.Finally through the workpiece fetching experiment, proves that the whole implementation process is simple and can meet the accuracy requirement of the grab.