When a tracked object is not occluded by others, it is easy to recognize it in the next frame, because its position has not changed excessively and can be predicted with techniques as the proposed tracking method based on Kalman predictions and LSAP association. However, if an object is in an occlusion situation, the methods which are only based on the object position or direction are inadequate, because, specially in long occlusions, the trajectory of an object can be substantially altered.