An approach of electronic stabilization based on binary image matching of color weight

Color weights of green(G) and red(R) are extracted from the color space RGB for color video image, and then are respectively converted to binary images, which are used in the following image matching. Methods of Large Diamond Search and Small Diamond Search are applied in template searching in the binary images with the values R and G. Method of block matching is used to obtain the motion vectors in every sub-region. Then global motion vectors can be got by the mean filtering between two inter-frames. The global motion vectors are then filtered by Kalman filter to remove the random jitter motion. A stable video image output is achieved by compensating the image frames and by using the global motion vectors filtered. According to the assessment standards of image stabilization, the effectiveness of the proposed algorithm is verified.