DEMO PAPER: VIDEO RETRIEVAL SYNOPSIS FOR MOVING OBJECTS

摘要

This paper demonstrates a novel retrieval synopsis system based on moving objects for surveillance video. With the popularization of digital video surveillance, massive data has been stored and the volume is still rising. How to utilize surveillance video effectively and efficiently is strategically important for practical applications. So as to improve the availability of video, intelligent applications, which contain object extraction, video indexing, video retrieval, and fast browsing, are performed with background modeling and retrieval synopsis. Specifically, retrieval synopsis offers three retrieval modes: playback retrieval mode, variable-fidelity retrieval mode, and attribute retrieval mode. That is, both integral synopsis video browsing and specific objects retrieval browsing can be executed on original video. As demos verified, the system can realize video retrieval and synopsis browsing in a flexible and efficient way.