FH-adaptive motion compensation for moving object detection in an active camera

An efficient camera motion compensation algorithm based on multi-layer decomposition is proposed for moving object detection in an active camera. Optical flow map obtained using the pyramid LK algorithm is divided into multi-layers by flow histogram decomposition. The camera motion model parameters of each layer are computed and used to compensate for the motion. The proposed algorithm has more accurate motion compensation and moving object detection results than the existing single-layer motion models. Also, the proposed algorithm is fast, so it can be applied to real time systems.