Electronic Image Stabilization based on Laplacian Bit Plane and Matching Block Classification

In this paper, an electronic image stabilization method based on Laplacian bit plane and matching block classification is proposed. The method uses Laplacian to extract the edge information of the image to construct bit plane. Weight is used to measure the feasibility of the vector becoming local motion vector and global motion vector. The matching result of each matching block and the global motion vector of the previous frames are combined to obtain the global motion of the current frame, which is used to compensate for frame motion to obtain a stable image. Finally, it is proved by experiment that the algorithm proposed in this paper requires a small amount of calculation and can eliminate the interference of small objects in the field of view. For CIF images, the algorithm can achieve real-time image processing, removing and reducing the jitter of the image. Compared with traditional bit plane matching electronic image stabilization methods, this algorithm lays stress on the full utilization of gray information and the adaptability of the algorithm when there is a small foreground object in the field of view.