Motion estimation with histogram distribution for visual surveillance

In this paper, we suggest the issues for detecting and tracking the objects. First, we utilize the MoG (Mixture of Gaussian) method to model the background for segmenting the pixels of background. Than, we can calculate the foreground (moving object) pixels using difference between background model and current frame. In order to get more accurate foreground, we recommended a method which is combine HSV and gradient distribution for removing the shadows. For objects tracker, we used approach that incorporates the Kalman filter estimation with histogram information. The idea of proposed method is calculating the motion estimation with colour histogram corresponding to the detected objects that we want to track. Finally, we proved the performance of the proposed algorithm.