An Object Surveillance System Based On Binocular Vision Guided By Fish-eye Camera

In the field of intelligent monitoring,the fisheye camera has a larger field of view than the traditional camera.However,due to the limitation of its resolution and the distortion of the fisheye image,the detail information of the target is hard to observe.To overcome this problem,binocular tracking system guided by fisheye camera is built in our paper.The system is designed to combine the advantages of fisheye lens cameras and traditional cameras with a fisheye camera as a guidance system for panoramic surveillance and a binocular camera as a tracking system for target tracking and distance measurement to achieve ultra-wide-angle,high-precision intelligence monitor.The main contents include:1.We use a change detection algorithm to detect the target based on local adaptive sensitivity(ie,Subsense),which uses the local binary similarity pattern descriptors(LBSP)and color intensities(RGB)to characterize pixel representations.Then,we use the obtained samples of pixel models for background modeling.In order to improve the flexibility of the system,a feedback mechanism is used to dynamically adjust the parameters like the foreground judgment threshold.Experimental results show that this method can identify the target of complex motion and meet the monitoring requirements of the system.2.Calculate the target space angle.Fitting the circle to obtain the main point coordinates of the fisheye lens image by the least square method.According to the fisheye lens imaging principle,the target image is designed and photographed to obtain a large number of image point angles and corresponding target point 3D space angles.We use these angles to obtain the calibration parameters by data fitting method.We use the calibration parameters to calculate the 3D space angle of the target corresponding to the target point in the image coordinate system,and then to obtain the turntable rotation angle.3.Through the overall system experiments to verify the binocular tracking system under the guidance of the fisheye camera,the effectiveness of the algorithm in this system is proved through the comparison and analysis of the specific experimental data,and it is pointed out that the system has not been implemented yet,which will provide guidance for the next study.The overall experimental results show that the change detection method adopted in this paper can accurately detect the moving target in the fisheye image,the angle calibration accuracy meets the system requirements,and the binocular camera can accurately detect the target distance.The intelligent monitoring system can reduce the influence of fish-eye images distortion on the detection results,has the advantages of wide field of view and high detection precision,and provides a detection method for intelligent monitoring technology.