Moving human tracking algorithm based on partial Hausdorff distance

In order to realize accurate and fast tracking moving human, a moving human tracking algorithm based on partial Hausdorff distance is presented. This algorithm adopts Gaussian mixture distribution to model background of each pixel and background subtraction to detect moving regions. Shadow elimination and after treatment of moving regions provide moving human regions. Kalman filter is used to predict the position of tracking target in the next frame. Partial Hausdorff distance calculation is used to realize exact match, and the purpose of accurate and efficient tracking moving human can be reached. The experimental results show that the algorithm can continually track moving human accurately and fast despite occlusions and other moving object interference.