Object Tracking in Video using Mean Shift Algorithm including Effect of AWGN channel

Abstract

One of the analytic ventures in object tracking is the tracking of fast-moving objects in arbitrary movement, mainly in the area of video vision applications. Thus a technique of mean shift (MS) algorithm in visual video tracking is put forward. In this suggested method, arbitrary motion and partial occlusion of an object can be managed due to its capacity in estimating the object position with modifying motion model. Although the techniques like particle filter (PF) is able to manage numerous hypotheses to manipulate clutters in background and short-term breakdown. However, on the other hand, it needs a huge number of particles to estimate the actual posterior of the target dynamics. Therefore, MS algorithm is employed to the sampling process of the PF to carry these particles in gradient ascent direction. As a result of this, a little sample size will be adequate to constitute the system dynamics precisely. The proposed algorithm is directed to track the moving object in arbitrary directions under altering states with reasonable computational time. The dissimilarity between the target model and the target candidates is expressed by a metric derived from the Bhattacharya Coefficient.