An Improved Object Tracking Algorithm Based on Camshift Combined with Active Contour and Kalman Filter

摘要

Camshift is an improved algorithm from Meanshift to track object using color probability distribution, but similar color from background and other objects can cause Camshift fail to track. This paper proposes an improved object tracking method to overcome the weakness of Camshift. Not only a weighted color histogram is generated to represent the object precisely, but also color probability distribution and moving probability distribution are integrated for Camshift to track efficiently. In addition, kalman filter, contour information and histogram matching are utilized to make the method for accurate tracking. Experimental results demonstrate the robustness and reliability of our improved tracking algorithm.