Research On Cooperative Reconnaissance And Control System Based On Multi-sensor

The multi-sensor cooperative reconnaissance system can detect the intrusion target around the clock in an unattended wild natural environment,and perform computer vision processing on the monitored target video stream to detect and track the moving target.Under the premise of limited energy,through reasonable coordination of nodes,it can not only ensure the multi-view and strong fault tolerance of the sensor network,but also improve system efficiency and extend the system life cycle.It has broad application prospects in the Internet of Things,military,environment,health,family and other commercial fields.Firstly,in order to solve the multi-sensor network cooperative scheduling problem,according to its multi-redundancy and energy-limited characteristics,the motion state of the target is estimated by establishing cluster table information for the sensor nodes and using the spherical simplex unscented Kalman filter(SSUKF),according to the above conditions a cluster generation method based on sensor sensing capability is proposed.The cluster head rotation method is developed by considering the residual energy of each sensor node and its tracking accuracy.The moving target is used as the excitation,and the surrounding nodes of the target are dynamically grouped to complete the tracking task,according to the above conditions a node cooperative scheduling and tracking strategy is proposed.The target tracking process is implemented by using the collaborative scheduling strategy simulation,which has better stability and robustness than the traditional method.Then,in order to use the information collected by the video sensor node to perform real-time reconnaissance tracking on the target,based on the lightweight deep learning target detection network SSD Mobilenetvl,by improving the network structure,using the more fine-grained feature map to participate in position regression and classification to integrate the network Contextual information,and by introducing the anti-residual module to improve the ability of the network to extract features,the detection accuracy is improved while ensuring the real-time detection speed,and the training verification is performed on the KITTI data set,and good results are obtained.Finally,in order to improve the friendliness of the multi-sensor collaborative control system,the popular GUI development framework of PyQt is used to complete the design of the system infrastructure and control platform based on the Python language.The development of video echo early warning,reconnaissance service data management,geographical situation joint display and other modules is realized by using the idea of modularization,the whole system has good scalability.