Research On Cooperative Reconnaissance Based On Radar And PTZ Camera Network

The reconnaissance of non-cooperative targets is an important foundation for digital and intelligent systems in many fields,including security and digital cities.Non-cooperative target reconnaissance often uses a variety of detection methods.Among them,cameras conform to human vision and have rich texture information;centimeter-wave radars can detect all-weather,medium and long distances and are cost-effective.These two detection sources have been widely used.Cooperative reconnaissance based on radar and vision has the characteristics of wide reconnaissance range,intuitive and convenient,and large amount of information,which can greatly improve the detection,identification and tracking capabilities of the system.This study focuses on the collaborative reconnaissance of radar and Pan-Tilt-Zoom(PTZ)camera network.The main contents are as follows:(1)Research PTZ camera internal parameters,camera pan-tilt and environment calibration.1)Internal parameter calibration in view of the problem that the Zhang's calibration method cannot be used correctly in the medium and long-distance zoom environment,the two-step calibration method is used to calculate the camera internal parameters,and the method is verified and compared with the Zhang's calibration method.2)Because PTZ camera has a rotating structure,this study proposes a trajectory model of the optical center calibrated for the camera pan-tilt.The input of the model is the camera rotation angle and the output is the optical center coordinates.3)The environment calibration studies how to transform the target from the radar to the coordinate system of each camera,which is used to guide the PTZ camera to detect the target.(2)Research on collaborative scheduling methods based on expert experience and Markov chain for radar and PTZ camera networks.The former is based on expert experience and analyzes the influence of other modules and higher-level instructions on the coordinated scheduling,so as to formulate the state transition criteria of the coordinated scheduling module;the latter simulates scenario in this study,regards the system state as a Markov decision process,and proposes to do as much as possible How high a priority target is detected as a positive revenue target,and taking into account the negative revenue brought by the camera rotation efficiency,the camera joint action that maximizes the total revenue is the best scheduling decision.(3)Design sub-kilometer-level infield and outfield collaborative reconnaissance experiments for the common and relay reconnaissance modes of single radar and dual cameras,and adopt a collaborative scheduling scheme based on expert experience to target non-cooperative pedestrians,cars,and unmanned vehicles.The aircraft target conducts a coordinated reconnaissance test.The method proposed in this study has been verified by simulation and measured data.The results show that: 1)The PTZ calibration method proposed in this study can availably enhance the accuracy of internal parameter calibration,which is more accurate than the result of Zhang's calibration method;2)The environment calibration method proposed in this study can complete the detection within the acceptable error range Sources' mutual guidance;3)For the Markov chain-based collaborative scheduling method,the revenue calculation method proposed in this study is more effective than the traditional method under simulation conditions;4)The collaborative scheduling method based on expert experience proposed in this study can complete two different modes of collaborative reconnaissance missions as required in sub-kilometer-level scenarios.