Satellite Video Stabilization Method Based on Global Motion Consistency

As a new earth observation technology, video satellite is of great significance for extracting and analyzing large-scale and deep-level dynamic information. Satellite video data will gradually become one of the important data sources for intelligent surveying and mapping application services. However, video satellites have problems such as random jitter, which leads to unstable video, and low matching accuracy between frames due to satellite video content containing moving objects. In this paper, a satellite video stabilization method based on global motion consistency constraints is proposed. The method first uses the lightweight ORB algorithm to extract and describe the feature points of each video frame and perform frame-by-frame matching. Secondly, global motion consistency is used to eliminate mismatch points and imprecise matching points. Then, the affine transformation matrix is obtained and linearized according to the matching point pairs. Finally, the linearized affine transformation matrix is used to resample frame by frame to obtain a stable video. To verify the effectiveness of the method in this paper, the video data of SkySat-1, Jilin-1 and Zhuhai-1 are selected for experiments. The experimental results show that the method proposed in this paper can overcome the inter-frame dynamic target interference and improve the inter-frame matching accuracy, to efficiently achieve stable satellite video output.