Research on Fast Identification and Location of Disordered Objects Based on Binocular Vision

Aiming at the problem of poor real-time performance of robot arm in identifying, positioning and grasping chaotic targets. A fast recognition and location algorithm of disordered targets based on binocular vision is designed. In this algorithm, an improved semi-global block matching algorithm is used to calculate the stereo matching of the target image, and the parallax map is output quickly by stereo matching of the image. Then combined with PCL library, the disordered target image is transformed into a point cloud image; Then point cloud processing and other operations are carried out; Finally, the point cloud image is segmented to screen out the first captured target, and the target point cloud is transferred from the camera coordinate system to the base coordinate system, and then the plane normal vector and plane center point coordinates of the target are obtained by plane fitting. So as to complete the rapid identification and location of disorderly targets. Experiments show that the improved algorithm can obviously reduce the time spent in stereo matching of disordered targets, and the calculation speed is increased by 30.7% compared with the traditional stereo matching.