Research And Implementation On Abnormal Human Behavior Detection In Monitoring System Of The Safe City

Moving object detection and tracking is a hot topic in computer vision,intelligent monitoring system and more and more people's attention.According to the relative motion between the camera and monitor background relationship,will target movement into the static context and dynamic context of movement.At present,the static background research of moving objects is relatively mature and widely used in all areas of society,whereas a dynamic target motion against the background of relatively complex theory and practical applications there are a lot of issues need to be resolved,this paper focuses on the dynamic context of the target detection and tracking of pedestrians,experimentally verifying the feasibility of this method.Firstly,reviewed the status of pedestrian detection and tracking,and analysis of pedestrian detection and tracking of various aspects of the problems which may arise in the process,combined with knowledge of digital image processing,analysis of target detection in the static context and dynamic context primary extraction method for both cases.Dynamic background of games by global motion of the interference caused by camera movement,need we to compensate for camera movement background.Point matching based on SIFT features,model of the affine parameter estimation,estimation process,presents a new interval matching on removal characteristics of Chinese and foreign interference.Object extraction using frame difference method to enhance the identification of target detection capability,well meet the testing needs.For target tracking of pedestrians,firstly introduced several traditional target tracking method,analyzes deficiencies of several algorithms,and apply the Camshift verified,that single feature is difficult to achieve its objectives in a complex environment described,prone to target loss situations.This multi-feature fusion method is proposed to improve the track count,for different target tracking under the environment needs,you can choose different features of fusion,improved tracking algorithm can be a good track to achieve improved the robustness of the algorithm.Dynamic target tracking in the context of having broad application prospects in intelligent vehicles,drones,special operations and other fields can be well used.This article focuses on achieving the goal of pedestrian detection and tracking on a mobile device is analyzed through experimental results verify the feasibility of the system,results show that the algorithms proposed in this paper to meet the requirements of pedestrian detection and tracking on a mobile device,have some practical value.