Semantic Analysis of Traffic Video Using Image Understanding

Video retrieval technology has always been the important application domain of image engineering. Aiming at the problem that low-level features cannot describe high-level semantic completely and accurately in the content-based video retrieval technology, this paper introduces the method s of image understanding and proposes a video semantic analysis framework using scene analysis for recognizing some semantic events in the traffic video. The experimental results show that the feasibility of the framework is good and the event recognition method based on scene analysis has good recognition effect for vehicle status in traffic video.