ROI Based Video Object Tracking Using Mean Kernel Profile of Histogram

Abstract: Object Tracking is one of the most challenging subjects in the field of computer vision, surveillance, traffic monitoring, video compression etc. The aim of object tracking is to find a moving object in a video frames sequence. Normally a video tracking system combines three stages of data tracking; object extraction, object recognition & tracking. We propose an approach for tracking object in a single frame in which a centre point of object is taken as focus component. The histogram profile based object representations are updated by changing kernels. To enhance localization of the tracked object some region bounding structure information is added to the method of tracking. This method is successfully adjusted with moving camera, Partial occlusions and changing scale and orientation of target have overcome the need of background subtraction making it more efficient. Some main applications are: surveillance application, control application and analysis application.

Keywords: Kernel, localization, mean, kalman, histogram.