Translation Based Estimation Technique to Handle Occlusion While Using Mean-Shift in Tracking

摘要

Detection of occlusion is a common problem in video surveillance field. When, moving objects overlap one another, an occlusion is created. It may happen very often in real life video. Many researchers have devoted their time to overcome that problem and have proposed many techniques. We propose here a blob translation based estimation technique to detect occlusion. We use the velocity, area and direction to do that.