Research And Implementation On Vision Based Abnormal Behavior Analysis Algorithm In Security System

Intelligent video surveillance technology plays an important role in protecting social public security, but it still has many limitations when analyzing complex actions. To figure out this problem, the camera calibration technique based on monocular vision is used to optimize the original video monitoring technology. This has theoretical significance and practical value.At first, the camera calibration technology is adopted to realize the coordinate transformation, and estimate the object’s size in image, which is of great significance for enhancing the universality of subsequent processing algorithm. As to object detecting, "model second match" method is used to obtain more complete object; Merge and filter objects referring to estimated image size. As to object tracking, Meanshift tracking window adaptive adjustment algorithm is used to reduce tracking lost; actual coordinate trajectory helps to improve precision of behavior analysis. For abnormal behavior analysis, three typical security abnormal behavior is studied, including area intrusion, following, hovering. Finally, abnormal behavior analysis prototype system is designed and implemented to verify the algorithm in this paper. The experimental results show that this algorithm can satisfy the requirements in security system.This project is based on security requirements, looking forward to contributing to the intelligent development of security system.