Intelligent Monitoring System Based On Video Detect And Its Key Techniques Research

Computer Vision is an important embranchment of Artificial Intelligence, and has been applied widely. The research about recognition and tracking of moving object become a hotspot recently. In the situation of the world that security issue has been paid unprecedented attention, intelligent monitoring system will be more widespread used in future. Intelligent monitoring techniques relate to numerous research domain, such as image processing, image analyze, machine vision, pattern recognition. It is a sort of leading integrated technique that intersect many subject and full of challenge.Firstly, this article will introduce simply the domestic and overseas research actuality about intelligent monitoring system and correlative techniques, and explain the common techniques in current intelligent monitoring system whose advantages and disadvantages is analyzed simultaneously. Based on these, the achievement is expatiated from three aspects as follow.1. Detection of moving object under fixed scene. The algorithm principium of background subtraction and time difference technique is explained in this article, and its advantages and disadvantages summarized and its applicable environment indicated by experiment.2. Recognition and tracking of weak moving object under complex background and weak contrast. Some assigned works have been done in this domain. At present, some methods such as infrared ray can be used in these scenes (the night scene for instance). In fact, the contrast of images obtained by these methods is not weak. This article puts forward a multiprocessing method to recognize weak object in condition of complex background and weak contrast.3. Elementary research of action analyze based on vector field model. It is a new technique of action analyze used in video monitor system. A dynamic vector field is obtained from video stream firstly. Some kind of event can be found out in video stream by variety of special parameters in the vector field to estimate if some abnormity arises.