C18. Modified CAMShift algorithm for adaptive window tracking

This paper introduces a proposed modified approach of an end-to-end technique for moving targets tracking. The tracking technique is processed on a real-time video stream from a single Pan, Tilt, and Zoom (PTZ) camera. The proposed approach is a prolongation of the Continuously Adaptive Mean Shift (CAMShift) algorithm applications. Sever variations in target shape, size and luminosity can be dealt better using this algorithm. Window difference technique is used to deal with the change in target shape and size. An estimator is used to deal with luminosity changes. A proper size of tracking window is built with minor extra computational overhead. Experimental results show the effectiveness of the proposed algorithm.