Object Tracking Using Improved CAMShiftAlgorithm Combined with Motion Segmentation

Continuously adaptive mean-shift(CAMShift) is anefficient and light-weight tracking algorithm developed based onmean-shift. While color based CAMShift is suitable for trackingtargets in simple cases, it fails to track objects in more complexsituations. In this paper we review our low cost extension toimprove the traditional CAMShift algorithm. Combining theoriginal algorithm with a motion segmentation phase, weproposed an improved CAMShift algorithm to cope withCAMShift`s tracking problems. We evaluated the efficiency ofour approach by comparing our tracking results with thetraditional algorithm`s results in several cases

Keywords:

Target tracking , CAMShift , motion segmentation , mean-shift , probability distribution image