Moving Target Detection Algorithm Based on Robust Background Motion Compensation

A new algorithm that affine model simulates cameras' complex motion in the scene was put forward,and then a precise motion compensation frame after robust estimation affine parameters through multi-scale motion estimation algorithm was obtained.After threshold the magnitude of optical flow between current frame and motion compensation frame,an initial target segmentation was obtained.At last,an intact target mask was detected through cluster and search fill algorithm.Experimental results show that this algorithm can eliminate global background motion very effectively because of cameras and segment target under moving cameras.