Detection And Tracking Of Moving Objects With Master-slave Camera Coordination In Large Scene

Detection and tracking of moving objects has always been one of the hot issues in the field of computer vision and image processing.Among them,the detection and tracking technology for moving targets in large scenes has wide application value and important theoretical research significance in military,security,transportation and other fields.However,the large scene monitoring has a wide field of view.If only a wide-angle camera with a fixed focal length is used,the target has a small proportion in the camera screen and cannot obtain the target high-definition image and details.PTZ control camera(Pan-Tilt-Zoom,PTZ)Although it is possible to obtain a clear image of the target by zooming,there is a problem that the field of view is small and it is easy to ignore the surrounding objects.Both cameras cannot meet the needs of large scene monitoring.In this thesis,according to the characteristics of large scene monitoring and its application requirements,combined with the advantages of the fixed wide-angle camera and the PTZ pan tilt control camera,the method of detecting and tracking the large scene moving target based on master-slave camera coordination is studied.Among them,the wide-angle camera is the main camera,which detects and tracks the moving objects in the entire area.The zoom PTZ camera is from the camera.The main research content of this article is as follows:(1)The method of detecting and tracking the moving target of the main camera under a large scene is studied.The main methods include: main camera image distortion correction,main camera moving target detection and tracking.Due to the distortion of the images captured by the wide-angle camera,this article first calibrated the main camera image.Then,this thesis combines the Vibe and three-frame difference algorithm to detect the moving target in the main camera image.Finally,based on the Hungarian algorithm,a preliminary match is achieved,and the Kalman filter is used for trajectory prediction to complete the target tracking of the main camera.Experiments show that this algorithm can detect and track all the moving objects in the scene at the same time.(2)The method of target detection and tracking from the camera is studied.This thesis compares the background characteristics of the PTZ image with the specific features of the camera image,compares the advantages and disadvantages of background compensation,traditional machine learning methods and deep learning methods in dealing with the target detection problems in the dynamic background,and uses a higher detection accuracy rate.And the real-time YOLOv2 method is better.Targeting the target color characteristics from the target characteristics of the camera as the target identification,Camshift and Distractor-Aware Tracking(DAT)were used to track the target,and combined with the YOLOv2 detection,the target was quickly and accurately detected and tracked from the camera image.The experimental results show that under the dynamic background,this method effectively achieves the target detection and has a good tracking effect.(3)For the cooperation problem between the master camera and the slave camera,the mapping relationship between the pixel coordinates of the master camera and the PTZ values(pitch angle,horizontal declination angle,and camera focal length)of the camera is studied.The algorithm uses the correspondence between master and slave cameras to track the same target and completes the establishment of the master-slave camera relationship.Based on this master-slave relationship,by comparing the PTZ value of the target coordinate in the main camera with the current PTZ value of the slave camera,the feedback information is used to control the slave camera action,which ensures that the target is in the middle position of the slave camera screen.(4)A large-scene master-slave camera coordinated moving target detection and tracking system is designed and implemented,which mainly includes four modules: main camera target detection tracking,camera target detection tracking,master-slave camera relationship determination,and master-slave control.First,establish the master-slave relationship between the wide-angle camera and the PTZ camera.then,detect and track the moving target in the main camera.then,based on the master-slave relationship constructed above,realize the control of the master camera to the slave camera.and finally,The target position is locked from the camera,detection and tracking are performed to solve the problem of detection and tracking of moving targets in a large scene.The system can not only achieve the overall monitoring of the entire region,but also can obtain the details of the target of interest,and provide technical support for subsequent target identification and behavior analysis.