移动机器人实时目标跟踪系统设计与实现

Design and implementation of real-time target tracking system for robot

针对单目视觉移动机器人目标跟踪的实时性和鲁棒性要求,提出了基于Kalman滤波器的改进Camshift算法检测和定位目标。将Kalman预测值作为目标初始位置,补偿摄像头和目标相对运动导致的目标在图像中的偏移。在系统"跟丢"后判断目标丢失的原因,根据原因自适应拓展搜索窗口作为Camshift算法的下一帧初始搜索窗口。为了验证改进算法的有效性,自主研制了一种应用该算法的履带式机器人实时目标跟踪系统。实验结果表明:该系统具有很好的鲁棒性和实时性。

Aiming at real-time and robustness requirement of target tracking requirement of mobile robot using monocular camera,an improved target tracking Camshift algorithm based on Kalman filter. To compensate the offset of the object appearring in the image sequences caused by camera and object relatively moving,use the Kalman predicted value as target initial position. When the object is lost from the tracking system,analyze the reasons at once. And accordingly,the size of search window is expanded adaptively,which act as the initial search window of next frame of Camshift algorithm. An real-time target tracking system for robot is developed to verify the effectiveness of the proposed algorithm. Experimental results show that the target tracking system using the proposed algorithm is capable of good real-time characteristic,and robustness.