Research On Space Target Detection Technology In Space-based Wide Area Surveillance System

With the development of science and technology and the increasing attention to the space environment of various countries,the exploration of space has continued to increase,and more and more spacecraft are sent into space.The resulting space environmental security issues cannot be ignored.Space surveillance is of great significance for maintaining the safety of space resources and avoiding space target collisions.It also provides necessary information support for spacecraft rendezvous,on-orbit maintenance,and space debris cleaning tasks.Among them,space target detection is the core technology of the space surveillance system,so it is of practical significance and application value to carry out in-depth research on this technology.In this paper,space-based wide area surveillance system is used to detect space targets.Compared with ground-based surveillance,the system has the advantages of not being restricted by light and weather,wide detection coverage,and flexible.However,at the same time,there are many difficulties in the implementation of space-based surveillance.Firstly,there are a large number of stars in the field of view,and their imaging is very similar to the space target,which is prone to false alarms and interferes with the detection of the space target.Secondly,the image may contain multiple space targets with different orbital heights,and their imaging size,energy,motion speed and direction are different,and the target trajectory may appear discontinuous,nonlinear,etc.,which greatly increases the difficulty of space target detection.Finally,the amount of image data is large.In hardware implementation,how to ensure the processing effect of the on-orbit image processing algorithm while saving hardware resources and improving the real-time performance of the system are also problems that we need to solve.The above problems bring many challenges to space target detection and implementation.Therefore,this paper focuses on three aspects: the removal of background noise and stars in space-based image,the detection of space targets and on-orbit image processing algorithm hardware rapid implementation,mainly including the following content.(1)Research on star image preprocessing technologyFirst,we analyze the imaging characteristics of star image background,estimate and remove the non-uniform background.After that,an improved adaptive threshold segmentation algorithm based on image line calculation is proposed,which can estimate the mean and standard deviation of background noise more accurately,and at the same time achieve the purpose of reducing computational cost.In addition,an omnidirectional morphological filtering algorithm based on time-index image is proposed to effectively remove background stars and noise,so as to reduce the interference of background stars and noise on target detection in the preprocessing stage.(2)Research on space-based space target detection technologyOn the basis of removing most of the background stars and noise,in view of the existence of multiple space targets in the field of view,and the target motion speed and direction are different,the trajectory is nonlinear and discontinuous.Firstly,an effective state transition region prediction algorithm based on the relative inter frame motion distance of the target is adopted to predict the motion of multiple space targets separately,and at the same time,the requirement for the centroid positioning accuracy is reduced.Secondly,a target detection algorithm based on state transition sliding window is proposed,which can effectively detect the space targets with discontinuous trajectories(the target is lost in a single frame image).When the signal-to-noise ratio is equal to 3,the detection probability can reach more than 98%.In order to further improve the detection probability of space target,a space target detection algorithm based on state transition multistage hypothesis testing is proposed.This algorithm improves the search range and search method of space target trajectory candidate points,and solves the problem of target detection under complex conditions.The experimental results show that when the signal-to-noise ratio is greater than or equal to 3,all space targets in the simulation image data set can be effectively detected,and the detection probability can reach 100%;when the signal-to-noise ratio is 1.5,the detection probability can still reach 92%.(3)Design of on-orbit image processing algorithm based on FPGABased on Virtex-4 series FPGA XC4VSX55 produced by Xilinx company,the on-orbit image processing algorithm is designed.Firstly,in the aspect of non-uniform background removal,one-dimensional structural elements are used to replace two-dimensional structural elements for morphological filtering,which facilitates the implementation of the algorithm on the hardware architecture while ensuring the filtering effect.Secondly,in the aspect of image threshold segmentation,the background noise mean value of the image after removing the non-uniform background can be ignored,and the multiple of the standard deviation is the segmentation threshold,so the threshold calculation formula can be simplified.After that,the corresponding hardware logic block diagram is designed for the algorithm,and the whole structure adopts the method of "global pipelining and local parallelization",making full use of the advantages of FPGA parallel execution,which greatly meets the real-time requirements of on-orbit image processing algorithm.