Research On Detection And Tracking Technology Of Human Motion For Video Surveillance

With the development of technology and the popularity of intelligent video surveillance,the need for detection and tracking of moving targets in the field of computer vision has become increasingly important.Although there have been plenty of methods to detect and track targets,the detection and tracking of moving targets has become a difficult task due to differences of human body imaging in video and the changes of background environment.The main work of this paper is to detect and track moving human body in static background.The key work contents are arranged as follows:In the detection of moving human motion,an improved traditional three-frame difference method for human contour detection is proposed.The current video frame background model is obtained by improving the mixed Gaussian background model,and then the current frame is obtained by background subtraction.Moving the target area,further dividing the moving target area into a non-dynamic area and a dynamic area,and then determining the obtained dynamic area,that is,an accurate moving target area,and updating the background area in real time with different background update rates;Meanwhile,Using the three-frame difference method to calculate the difference between the current frame of video and the previous frame and the next frame,the ‘AND' is performed on the difference result,and the obtained moving target region is compared with the result of the ‘AND' operation.The ‘OR' operation and the morphological correlation processing of the results of the operation yields a final smooth contour region..In the field of sports body tracking,several mainstream target tracking techniques are first described,including target tracking techniques based on regions,shapes and models.At the same time,based on the traditional tracking method,combined with Kalman filtering and Meanshift tracking algorithm,incorporating the improved idea,using geometric features to describe the observation model of the human target,adopting the adaptive fusion strategy,that is,according to the feature information in the tracking process According to the credibility of different environments,the weights in the fusion process are adjusted.The kalman filter is used to predict the target path,and then the mean shift algorithm is used to locate the moving target position information,iteratively judges,and the tracking frame is used to identify the target.At the same time,occlusion judgment is used to achieve robust tracking of moving human targets..In this paper,the research mainly uses Matlab tools and HD video to simulate and simulate the algorithm,achieving the detection and effective tracking of the moving human body,compares the performance and accuracy of the improved algorithm,and determines the effectiveness of the detection and tracking algorithm.