OCCLUSION HANDLING IN OBJECT TRACKING

Abstract

Object tracking is an essential component of an intelligent video surveillance system. Accurate and real-time object tracking will greatly improve the performance of object recognition, activity analysis and high-level event understanding. However, strength of kalman filter can be used to track object in adverse situation. Integrating a kalman filter within a standard tracking system allows the kalman filter is to use progressively updated features and aids in main training identity of the tracked object, and provides tracking system with an effective means. This thesis shows the possibilities of detecting the locations of unattended luggage’s and object or a ball in occluded scenes in which the positions of these objects before occlusions are taken and then a estimate of that position is given as a primary input in kalman filter which is the main sole of this thesis. In this thesis, we tried different videos with fixed camera with a single object and multiple objects to see it is able to detect objects. This paper presents a robust tracking approach to handle challenges such as occlusion.