Research And Implementation Of Background Modeling Method Under Complex Scene

As the basis of numerous computer vision,moving target detection technology has become a key research issue of the intelligent monitoring system, target recognition, human-computer interaction and scene analysis involved in image and video analysis and understanding. The task of moving target detection is to extract moving object of interest as complete as possible from the complex scenes, which depends on the accuracy of background model. Therefore, the accuracy of the background model has an important impact on subsequent judgment, such as foreground detection, classification, tracking and behavior analysis to understand etc. In other words, it largely determines the overall performance of the video surveillance system.This subject mainly studies background modeling techniques are used to detect moving target in complex scenes. In view of engineering application, this paper in-depth analyzes several classic background modeling methods selectively, and summarizes the advantages and disadvantages of each algorithm depended on the modeling theory and the experimental results. Ultimately, the block background modeling method based on the combination of higher moments and canny edge operator is designed. Aiming at the situation of slow moving object and stop for a long time in the video scene, it uses inter-frame continuity testing plan to make the background modeling algorithm more reliable. Experiments show that, under the premise to ensure the accurate detection of moving targets, the algorithm can reduce the time and space complexity of background model reconstruction, and improve the operating efficiency of the system.