Research On Close-loop Simulation Of Staring Image And Detection Of Infrared Small Target

The low-orbit gaze imaging platform has the advantages of high resolution,strong mobility and low launch cost,but the research and application of low-orbit gaze satellite is still in the stage of demonstration at present,and both the on-orbit target detection scheme and the performance index of the satellite platform are lack of experimental verification.This paper aims at extracting and confirming the small infrared targets under the low-orbit moving platform and surface array gaze imaging system.Firstly,the imaging characteristics and target background characteristics of the platform under the current working mode are analyzed,and then the infrared imaging simulation software is designed to obtain the test image.The main works and ach ievements are as follows:(1)Analysis and Simulation of Staring Imaging Characteristics of Low-orbit Satellite Platform.The imaging characteristics caused by the motion of the satellite platform are analyzed.The problems include image shift and blur caus ed by the motion of the platform,camera distortion caused by the motion of the platform in integral time.The main target and background model are determined by using the background of this project.The simulation function of infrared gaze image of moving platform is realized,and it has the simulation ability of different signal-hybrid ratio and different scale in typical mountain range,ocean and town scene.(2)Adaptive extraction algorithm for suspected target based on clutter suppression.The weak target in the infrared image usually occupies less number of pixels in the detector image surface and does not have significant morphological characteristics,so the method of detecting by matching the target template is not suitable for the background of this project.the key problem of background suppression is how to estimate the scene clutter distribution.to solve this problem,we need to obtain enough a priori of the scene distribution,and design a background estimation algorithm with high robustness and accuracy.At the same time,it is necessary to design an adaptive extraction function in the candidate target extraction stage to complete the extraction of candidate targets in complex scenes,and the algorithm should have the ability to suppress false alarm such as scene distortion and noise.The target extraction method based on clutter suppression can accurately extract suspected targets in infrared scene,and the detection rate of targets with signal-to-hybrid ratio greater than 5 and diameter of 2-3 pixels is over 98%,and the virtual scene rate in the detection results is about 10-5.(3)Target recognition algorithm based on multi-frame trajectory association.In the process of infrared detection of small and weak targets,because of the lack of geometric scale characteristics and low signal-to-noise ratio,only the spatial characteristics are used to detect them,and the detection performance is often poor,which can not meet the high detection rate and alarm rate.Therefore,it is necessary to combin e multi-frame time-domain method to detect the association of targets,so as to expand the detection scale and improve the detection performance.The key research contents in this process are the management strategy of trajectory database,the matching of low orbit imaging characteristics,and the establishment of determination calculation model.Correlation detection utilizes the correlation of the target in time domain,such as energy continuity,motion continuity,and trajectory correlation.All the cand idate targets in each time scene are associated with more than one space trajectory,and the high determination target is selected by judging whether the trajectory characteristic accords with the target prior characteristic.When the detection of small infrared targets(snr? 3,2×2 pixels)is carried out,the false alarm rate of target detection is reduced to about 10-6 without affecting the detection rate,and the accuracy of trajectory correlation is over 95%.