Moving Vehicle Electronic Image Stabilization Study

The image stabilization technology has gradually become an important part of video processing system in recent years, which could obtain stable video output by virtual of the elimination of the random jitter component. Video image stabilization generally can be divided into local and back-end image stabilization. The local image stabilization system can better ensure the requirements of image quality and real-time processing in comparison with the back-end image stabilization. From the view of video communication, local image stabilization system is more conducive to compress the video source compose of the stable image sequences due to the fact that it is easily implemented to achieve a smaller data stream at the same image quality, save the width of channel, and improve the efficiency of the communication system. In addition, compared to optical and mechanical image stabilization system, the electronic image stabilization system has been widely used due to its advantages such as small size, reasonable cost and low power consumption.In the thesis, the research background, significance and the development of image stabilization are introduced, and then the conventional model of image stabilization is described, the involved algorithms including blocks matching, bit-plane matching, characteristic points matching, features matching, optical flow algorithm, as well as low-pass filtering, B-spline curve motion filtering, K-means clustering and recursive Kalman filtering are represented. An image stabilization system is proposed. In the proposed stabilization system, the video source to be stabilized is converted to the bit-plane representative, and then the local motion vectors and global motion vectors are obtained via the block matching procedure. The real-time amended Kalman filter is applied to the results to implement the motion filtering and image compensation. The filtering parameters are adjusted by comparing the jitter parameters sampled by the sensors with those parameters calculated by the block matching procedure to improve the Kalman filter's performanceThe performance of the proposed system is evaluated using some video image sequences. The experimental results indicate that the proposed system yields good performance in terms of PSNR (peak signal-to-noise ration).