Research On Moving Pedestrian Analysis In Intelligent Video Surveillance System

With the development of computer and communication technology, video surveillance plays an extremely important role in daily security. The traditional video surveillance mode is mainly to collect video data stored in the monitoring center of the server. It has only a single view function, which can not meet the needs of users. This method of relying on artificial analysis of moving pedestrians not only lacks the function of timely warning of abnormal information, but also has low accuracy. Based on the research of moving target detection, pedestrian recognition and pedestrian abnormal behavior detection algorithm, this paper designs an intelligent video monitoring system, which can alarm the abnormal behavior of pedestrian movement.Firstly, an improved moving target detection algorithm is proposed to improve the moving target detection. In order to overcome the shortcomings of motion fading, light disturbance and poor background segmentation in the background, this paper combines the image block and mean method to establish the background model, and uses the image block and the current frame adaptive weight to update the background model and adopt adaptive threshold segmentation target. The experimental results show that the method is fast and can detect the moving target accurately.Secondly, a pedestrian detection algorithm based on multi-feature Fusion is proposed for the problem of low single-object pedestrian recognition. This method combines the improved multi -scale HOG features and CSSF features, which accurately describes the pedestrian local characteristics and global characteristics. An Adaboost strong classifier was designed for pedestrian detection. Experiments on the INRTA pedestrian pool show that this method has greatly improved pedestrian detection accuracy.Then, according to the problems existing in pedestrian abnormal behavior detection, On the basis of target tracking, the multi-feature extraction of pedestrian shape and trajectory characteristics is carried out, which can fully describe the pedestrian behavior information. Then,use the prior knowledge to quantify the above characteristics, and use it to detect pedestrian abnormal behavior. The experimental results show that this method can effectively detect pedestrian abnormal behavior, including abnormal fall, abnormal running and wandering.Finally, on the basis of the above algorithm, the requirement of intelligent video surveillance system is analyzed and designed. The development of intelligent video system is based on the VisualStudio, OpenCV, Android, JavaWeb, cloud server and the other technologies.Therefore, the video surveillance system becomes digital, intelligent and mobile. The experimental results show that the system has good real-time performance and accuracy, which can satisfy the user remote monitoring, and timely grasp of pedestrian abnormal behavior information function requirements.