Object Tracking using Generalized Gradient Vector Flow


The aim of an object tracker is to generate the trajectory of an object over timeby locating its position in every frame of the video. In this research, we present anobject contour tracking approach using Generalized Gradient Vector Flow(GGVF). GGVF active contour, or snake, is a dynamic curve that moves within animage domain to capture desired image features. Mostly, GGVF is not sensitive toinitial conditions and converges to the optimal contour. Given an initial contournear the object in the first video frame, GGVF can iteratively converge to anoptimal object boundary. In each video frame thereafter, the resulting contour inthe previous video frame is taken as initialization so the algorithm consists of twosteps. In the first step, the initial contour is applied to the desired object in firstvideo frame. The resulting contour is taken as initialization of the second step,which applies GGVF to current video frame. To evaluate the tracking performance,we applied the algorithm to several real world video sequences. Experimentalresults are provided.