A novel real-time digital video stabilization algorithm based on the improved diamond search and modified Kalman filter

Aiming at the real-time video stabilization under the potential panning motion, this paper presents a robust adaptive video stabilization algorithm based on an improved diamond search and a modified Kalman filter. On the one hand, by introducing Hooke Jeeves algorithm into the traditional diamond-search-based fast block matching method, the efficiency of motion estimation is improved. On the other hand, a modified Kalman filter is proposed to provide an adaptive motion compensation of quivering image sequences while the panning motion is preserved. Experimental results demonstrate that the proposed algorithm provides robust and adaptive compensation for image sequences even if fast panning motion exists.