双目视觉运动目标跟踪定位系统的设计

针对运动目标在被遮挡时跟踪丢失问题,采用双目视觉对运动目标进行跟踪定位.首先,利用背景差分法实现目标检测;然后,利用Kalman滤波器改进的CamShift算法与FAST角点检测算法相结合,通过缩小角点检测的范围,提高预测的准确性和跟踪速度,同时有效解决了目标跟踪丢失问题;最后,通过双目立体视觉视差原理求出目标的三维坐标,实现对目标的定位.实验结果表明,该系统有效地解决了目标跟踪丢失问题,且算法实时性良好,有利于工业上使用机器人对运动目标的精确抓取.

Design of Tracking and Positioning System of Moving Object on Binocular Vision

To solve the problem that loss of object information under occlusion causes the failure of tracking,binocular vision is used to track and locate the moving object.Firstly,the background difference method was used to achieve target detection.Secondly,this method that FAST (feature from accelerated segment test) corner detection algorithm that was combined with the optimized CamShift algorithm based on Kalman filtering was employed to reduce the search scope to improve the prediction accuracy and tracking speed,and effectively solve the problem of object loss.Finally,the three-dimensional coordinates of moving target were calculated with binocular stereo vision parallax principle,and the moving target could be located.The experimental results show that this system effectively solves the problem of targets tracking loss,and the algorithm has good realtime performance,which is helpful for robots to grab the moving targets accurately in the industry.