Research On Image Processing System Of Remote Control Reconnaissance Instrument

Remote control reconnaissance instrument is a reconnaissance equipment suitable for combat and counter-terrorism.In some dangerous areas,360-degree video information in the reconnaissance area can be acquired in real time within a certain distance by operating a remote control reconnaissance instrument.First,the development and current status of the remote control reconnaissance instrument were investigated,and an improvement plan based on the requirements for the designed remote control reconnaissance instrument is proposed.The original remote control reconnaissance instrument can only display the video information collected from a camera.Therefore,it cannot achieve 360-degree reconnaissance without blind spots.Its code is optimized,and its display interface is redesigned to realize the expectation that the interface simultaneously displays video information collected from four cameras.The original remote control reconnaissance instrument was only able to capture and display video,but it did not properly process the video.Therefore,this paper designs an image processing system based on the remote control reconnaissance instrument,which gives the remote control reconnaissance instrument the ability to detect and track targets,and realizes the intelligence of the remote control reconnaissance instrument.The development and current status of algorithms for detecting and tracking targets are investigated,and the algorithms for target detection and tracking are improved.An adaptive algorithm that combines frame differential method and background differential method for detecting target and an algorithm based on Kalman filter and mean shift for tracking target are proposed.First is the part of detecting target.Single Gaussian Method is used to build the background model,but there is a gap in the target image detected by this method.The image obtained by three-frame differential method is used to make up for the gap,so that the image of the detected target is more complete.On this basis,the learning rate of Single Gaussian Method is updated according to the frames of video image.The algorithm is used for experimental testing,and the detection effect of the algorithm is still good.Second is the part of tracking target.The algorithm is to use the algorithm based on Kalman filter to predict the next frame based on the observed values.Then it performs an iterative operation based on the mean shift in the predicted area to quickly calculate the position of the target and completes tracking of the target.After several experimental tests,the results show that the performance of the image processing system for detecting and tracking targets is stable,which meets the functional requirements of remote control reconnaissance instrument.