Research Of Human Skeleton Model And Abnormal Behavior Analysis Based On Video

Intelligent analysis of abnormal behavior research involves artificial intelligence, computer vision and pattern recognition, which provides supports for basic researches such as military targets locking and statuses judgment, real- time analysis of the behavior in extreme hazardous environments and security monitoring in public areas. In recent years, a large number of scholars dedicate to research related fields and try to make intelligent behavior analysis more accuracy and real-time when it applies to other areas.This paper proposes the research ideas for abnormal behavior in the video environment. There are mainly three parts of research contents, the first one is particle filter moving targets tracking algorithm based on color features. It uses color features because of color features having good robustness on non-rigid deformation, partial occlusion and target plane rotation. The second one is behavior analysis based on the human skeleton and human parametric modeling. In this part, we refine the target contour to obtain the skeleton image, and next employ a modified human skeleton key point positioning algorithm to extract the key points and establish the human parameter model. It converts the problem of human behavior analysis to that of solving parameters, reduces a large amount of data when calculates and and improves the efficiency of behavior analysis. We focus on human behavior analysis using parametric method in this part and mainly include determination for the parameterization behavioral status and parameter matching between two behavior states. The last one is abnormal behavior analysis based on fuzzy theory. It calculates exception of a whole people according to angles changes in the limbs and torso by fuzzy function and identify human behaviors at last.The further research of the subject will play an important role in promoting intelligent content search and early warning about abnormal behavior in video surveillance, and lay the theoretical foundation for building intelligent video surveillance system and retrieval of professional video content. Let intelligent behavior analysis can be applied to common everyday life earlier.