Implementation of Modified Mean-shift Tracking Algorithm for Occlusion Handling

Object tracking is critical and difficult task when it comes to unmanned air vehicles and traffic surveillance. The major challenge in object tracking is occlusion handling either partial occlusion or full occlusion. In this paper modified mean-shift tracking algorithm is proposed to tackle the problem of full occlusion. Mean-shift object tracking algorithm uses the color information to represent the target and to localize it in next frame. So when the object gets occluded with other object having similar colors, mean-shift tracking algorithm easily lost the target. In this modified mean-shift algorithm implementation, traditional mean-shift algorithm is lumped with the motion information associated with the spatial information of the moving object. Spatial information was exploited to handle the full occlusions present in the video. The object moves from one pixel to other in two consecutive frames its information about pixel index was stored in new variables. This information was used to capture the correct object when it reappears in the video, after the occlusion. Many videos were used to test the proposed tracking algorithm. Two examples were presented in this paper, which successfully cope with the partial occlusions, full occlusions and full occlusions when both the objects have exactly same colors. [Baber Khan, Ahmad Khalil Khan, Gulistan Raja, Muhammad Haroon Yousaf. Implementation of Modified Mean-shift Tracking Algorithm for Occlusion Handling. Life Sci J 2013; 10(11s): 337-342]. (ISSN: 1097-8135). http://www.lifesciencesite.com 62