Combine Kalman filter and particle filter to improve color tracking algorithm

In machine vision, color tracking is a well known problem. The Kalman filter or particle filter are often used to build color tracking algorithms. The Kalman filter is good in tracking a linear system, but it often misses the object when the object changes its direction suddenly. In this case, the particle filter is used but it fails easily when the object moves too fast. This paper presents another method to track a rigid object. Based on combining the Kalman filter and particle filter, this method increases the accuracy and speed of the color tracking algorithm.