A multi-mode image tracking system based on distributed fusion

In this paper, a multi-mode image tracking system was designed, which uses the distributed structure to fuse the tracking results of these modes. Because of noise in the image, the presence of occlusion, and deformation of the tracked target, some tracking algorithms may fail. The aim of the paper is to merge the advantages of all tracking algorithms to obtain robust behaviour. There are three tracking models applied in this system, based on region, B-snake, and distinguished points. Each of these three modes contributes to the fusion center, and communicates with other models. Results show that this system could overcome common problems in tracking.