Video Stabilization Algorithm Based on Video Object Segmentation

摘要

The goal of video stabilization mainly is to solve the blurred video caused by the unwanted camera movements. Based on the concept of video object which is proposed by MPEG-4, we apply video object extraction algorithm, simplified SIFT feature point extraction algorithm and motion compensation algorithm to process the video data captured from the fire scene and turn it to a clearer piece of video. Specially, this article proposes an algorithm, which can treat the different video objects respectively based on their value of information and reduce the time wasted on the background region effectively.