On vision servo tracking and Control based on multi-cue adaptive integration

A vision servo tracking algorithm based on adaptive multi-cue integration mechanism is proposed. The RGB color cue and edge cue are utilized to represent the target and then they are joined to the similarity function with which mean shift algorithm is used to find out the optimal location by iterative computation. The feature weight is used to adjust adaptively in the tracking procedure, Bhattacharyya coefficient is used as criterions for selective sub-model update. Experimental results show that the algorithm can achieve stable tracking performance in dynamical scene with the appearance of the target changing. Compared with tracker using single cue, the algorithm in a complex environment can real-timely track the target more accurate and robustly, and achieve the real time performance of vision servo tracking control.