Digital image stabilization with motion estimation based on binocular ranging and sensor

The stability of the output image sequences is vital for camera system. In order to stabilize video image sequences, a new digital image stabilization system with motion estimation based on binocular ranging and sensor is proposed. According to the imaging model of camera, the estimation of image motion vector is based on the distance of objects and the attitude information of camera. Binocular vision system is used to measure the distance of objects. The attitude information of camera is obtained by a high-precision attitude sensor. The motion compensation can be realized through affine transform. Experiments show that the system can process image sequences at the rate of 17 frames per second with good robustness and the stabilization precision is less than two pixels.