A variable search space (VSS) approachaccording to the color feature combined with point featurefor object tracking is presented. Mean shift is a wellestablishedand fundamental algorithm that works on thebasis of color probability distributions, and is robust togiven color targets. As it solely depends upon back projectedprobabilities, it may miss the targets because of illuminationand noise. To overcome the flaw, we proposes VSSalgorithm based on the color and robust feature of thedetected object. The proposed algorithm can solve theproblem that the color of the detected object is similar to thebackground, and achieve better real-time tracking due tochange the search window’s size. Experimental workdemonstrates that the presented method is robust andcomputationally effective.
展开▼