Research On Target Location Technology Based On CCTV Sensor Network

In the Inland and Maritime Vessel Traffic Service(VTS),the acquisition of position and motion status information of ship targets mainly depends on radar and radar data processing and thie Automatic Identification System(AIS),but The AIS subsystem is a passive information 'system,and the Closed Circuit Television(CCTV)monitoring system serves as an active information perception system.However,because the CCTV system does not have the target's spatial positioning and automatic tracking functions,its application in ship dynamic management is greatly restricted,and it can only be used as a radar auxiliary means to observe the water traffic scene.If the CCTV system is given the function of spatial positioning and track acquisition of the target,CCTV can replace radar within its surveillance range to achieve a superior water traffic briefing function than radar,and the realization of this function can obtain a higher Benefit investment ratio.This subject is devoted to the research of the CCTV spatial positioning method,and the estimation of the target spatial position based on CCTV is the basis for the realization of the target track estimation.The requeirments of the the target positioning and tracking system for the CCTV sensor network,the main research content and innovations are as follows:1)Designed a basic system of target positioning based on CCTV sensor network,and further expanded the system,designed a complete target positioning system based on CCTV sensor network,and designed to give the overall architecture of the system and target tracking Workflow.2)Aiming at the problem of low detection rate of small targets in the actual application scenarios of this subject,a small target detection algorithm based on improved YOLOv3 is proposed,a new feature fusion target detection layer is constructed,and the predicted value is output at a smaller scale.Through comparative experiments,it is proved that the improved YOLOv3 has a slight improvement in small target detection accuracy and recall rate compared to YOLOv3.3)According to the actual 'application scenario of CCTV sensor network,considering the advantages of monocular and binocular visual positioning and considering the situation of joint deployment,a monocular positioning algorithm based on visual imaging model and a binocular stereo positioning algorithm based on angle measurement are designed respectively.Simulation experiments prove the effectiveness of the two positioning algorithms.4)Aiming at the camera node pair scheduling problem involved in the binocular vision positioning algorithm,a Geometric Dilution of Precision(GDOP)node scheduling decision mechanism is established,and a node scheduling algorithm based on position prediction is proposed.Simulation experiments show that the algorithm has superior performance in terms of average GDOP and tracking error.In short,the purpose of the small target detection algorithm studied in this paper is to provide the necessary parameters to the target space positioning algorithm.The target space positioning algorithm is to be able to effectively support the continuous track tracking of the target.And the reasonable CCTV sensor network node scheduling strategy can better complete the spatial positioning measurement and continuous track tracking of the target.