Deep Compression: A Compression Technology for Apron Surveillance Video

Abstract

This paper presents a deep compression method that uses some object detection methods to separate moving and stationary objects from a real frame in an apron surveillance video. The extracted object image, the corresponding position information and the background image are stored in a linked list. When the video is decompressed, the extracted object images are restored to the background image according to the corresponding information, and the overall adjustment, such as illumination, is performed according to the stored information. Finally, a video with high similarity to the original video is generated. This video compression method greatly reduces the required storage space without destroying the original video information.