Study On Human Abnormal Behavior Detection Based On Video Sequence

With the development of computer and video monitoring, video processing technology hasbeen widely used in the national life. Based on human abnormal behavior detection of the videosequence can automatically analysis the even that occur in the monitoring scene, if there areabnormal behaviors in the monitoring scene, so it will be able to report to the police timely. So thescheme can not only improve the efficiency of monitoring, and save much lost, but also has thegood applicational value.By using the technologies such as the computer graphics technology,this paper realizes theautomatic analysis of target detection and tracking, then on this basis, according to the behavior ofthe target to judge whether the abnormal behavior.The three core module of this paper mainly is themoving target detection, tracking and human abnormal behavior detection. The completed researchwork is the following three aspects:1. In the motion target detection, this paper completes thecombination of interframe difference method and background reduction division to detect themoving target, and puts forward based on the selective background update of the area, and thenshadow detection and removal of the collected target.2. In the moving target tracking, this papercompletes the regional tracking algorithm of based on the feature matching, this algorithm uses theoverlap area of the previous frame targets and the current frame prospect pieces as the maincharacteristics, the target color histogram and position as the auxiliary features to track the human.The core of the algorithm analyzes mainly the relationship between the prospect block and thetarget.3. In the human abnormal behavior detection, according to the different abnormalbehaviors,this paper extracts the different characteristics and takes different methods to test, andthen analyzes primarily the line detection of the eight directions,wandering and faint detection.A large number of experiments show that this approach has a good adaptability and lowalgorithm complexity and so on. The detection scheme is able to detect the abnormal behavior ofhuman body, so it can be widely used in residential area, square, subway and so on.