Research On Space Based Small Extended Target Tracking Technology

The space-based infrared optical surveillance system uses the constellation of optical sensors deployed on multiple orbital platforms to detect,track,identify and forecast important events such as ballistic missiles and space launches in real time.Achieving immediate detection of potentially significant threats to homeland security will help rapid response and timely decision-making,enhancing our strategic defense capabilities.In this application background,space-based infrared optical monitoring system is bound to face the threat test of many kinds and model targets,so the detection system and the information processing technology involved in the whole system have a decisive influence on the system effectiveness.At present,the space-based part of the world's mainstream space-based infrared optical monitoring system is composed of multi-track platform equipped with multi-band optical sensors,using the dual/multi-star line of sight cross-positioning system,to complete the three-dimensional positioning tracking of space targets.In order to achieve the more ideal coverage of the world,the constellation system required is very large in scale,the cost ratio is very low,and faced with multi-target mid-range maneuvering scene,the satellite resources in local airspace are seriously inadequate,and the system has limitations that cannot be remedied.It is very urgent to study and explore a new detection system with stronger system performance and higher efficiency ratio.To this end,this paper discusses a two-mode detection system based on the main/passive combination,which combines the laser's one-dimensional ranging capability with the infrared two-dimensional angular measurement capability to achieve the rapid positioning and tracking of single-star single/multi-targets.In the information processing subsystem of the main/passive joint detection system,the information processing of the infrared angle measurement subsystem involves the problem of different scale,single/multi-target tracking,ballistic parameter estimation,and the information processing of laser ranging subsystem involves the real-time extraction of laser echo data and the problem of beam-to-multi-target traversal strategy.In addition,in order to achieve stereoscopic tracking of the target,it also involves the fusion of heterogeneous sensor data.These issues are discussed in different chapters.The second chapter first analyzes the system system of infrared and laser joint detection system,designs the composition and information processing framework of the infrared laser joint system,analyzes the serious challenges and shortcomings of the traditional early warning system represented by the "space tracking and monitoring system" in the United States,and highlights the significant advantages of the laser infrared joint detection system.On the basis of the level of the existing laser development,the feasibility of the main passive joint detection is initially explored,the theoretical strength of laser echo is analyzed numerically,and the future research content is laid out.This paper analyzes the simulation method of the extended target face from object to image,and on this basis,the simulation gives a general and more concise surface expansion target simulation method,which provides the entry point for the expansion target of the study of complex forms in chapter three.This paper combs out the motion law of the middle-segment target under the inertia system and the imaging process of the medium-segment target infrared sensor,and simulates the motion mode of the infrared group target in the plane,and finally,constructs a typical group target simulation scene as the data input of the fourth chapter research content.The third chapter studies the expansion target tracking of different scales and forms,and puts forward different tracking methods for the scale extension target and the surface extension target.In view of the characteristics of rich information and stable form of surface expansion target,a method is proposed to estimate both the target's mass and the target form.This chapter introduces Gaussian process method to describe the target expansion pattern,establishes the measurement equation and motion equation adapted to the stereoscopic target,and uses the extended Kalman filter to realize the estimation of the target centulateal core tracking and expansion pattern.The simulation proves that the proposed algorithm can effectively track the expansion target mass and expansion pattern.In view of the characteristics of less information about the pattern of the spot expansion target,the focus is on the center-of-mass tracking of the spot expansion target.Because the measurement of the spot target contains amplitude value information,its centrofness can describe the actual location of the target more accurately than the center of the shape,for which this chapter deduces a plausible function that incorporates the amplitude information.A simple and effective method is proposed to divide and package the image measurement,and further,the detailed steps for the standard probability hypothesis density filtering by using the three-dimensional box are given.Through simulation,we have verified that the proposed filter has better performance in tracking the expansion target.The fourth chapter studies the tracking problems of the middle group target in the semi-mixed and distinguishable states,and realizes the multi-target tracking and trail output function.Group targets in a semi-mixed state present many,mixed and dense features on the surface,and with the change of sensor observation angle,the tracking difficulty lies in the difficulty of accurately distinguishing the number of targets within the group,and thus can not achieve the estimation of multi-target state and number.To this end,this chapter first introduces the super-resolution method to achieve the resolution of semi-mixed state groups,and secondly,in view of the large computational overhead and time-consuming introduced by the hyperresolution algorithm,the execution method reference box particle filtering of the tracking algorithm,the target state is described and the measurement is packaged in the form of box particles,and then the box particle PHD is then used to The execution method introduces the scene,and finally achieves the multi-target track output function by adding trace labels.The label box particle PHD filter guarantees tracking accuracy and significantly saves computational resources,while meeting the need for an early output trajectory while significantly reducing computational time.In view of the problem of target-intensive distribution and data correlation difficulty caused by field-of-view jitter in the target tracking of distinguished group targets,this chapter regards the group of multi-proximity targets in the group as a structural stability diagram,and each target is a graph node,which describes the relationship between multi-target location structures by introducing hypergraph theory.According to the relative characteristics of group structure,abandonthed the traditional logic of calculating the strength of correlation based on distance,and realized the multi-target data correlation between frames by using the supergraph matching method,avoiding the error correlation caused by relying only on distance to judge the correlation logic,and significantly improving the accuracy of tracking accuracy and trace output.Chapter 5 focuses on the information processing involved in active sensors and the fusion of primary/passive sensor data.Aiming at the characteristics of laser reactive far higher than infrared frame rate and low detection rate,this chapter combines the laser echo data with time domain accumulation,and the method of maintaining the accumulation cycle in line with the infrared frame rate can not only increase the number of laser echoes in a cycle,but also achieve the goal of heterogeneous sensor cycle synchronization.Because the laser echo is one-dimensional distance information,and the traditional particle is a two-dimensional point,this chapter designs a line-type particle that can accurately adapt to the laser back portability,and combines it with the traditional particle filtering algorithm,effectively realizes the real-time detection and extraction of laser ranging information.In addition,the proposed line particle filter is verified by using simulation data and measured data from different ground-based laser ranging devices,and good results are obtained.Since laser spots usually cover only a single target,in multi-target tracking scenarios,an optimal pointing strategy is required based on the multi-target distribution location of the infrared image surface.This chapter models the traversal sequence problem of multi-target spot pointing to the face as a traveler's problem,uses the genetic algorithm to obtain the shortest solution for traversal,that is,points to the strategy,and uses the extracted ranging information and infrared two-dimensional angular information to implement the Carrman filtering algorithm for each target in turn.Combined with high-precision ranging information,the target 3D state tracking accuracy will be significantly improved.Two more active sensors on multi-target traversal methods,simulation of the two traversal mode of the target group tracking accuracy,analysis of the performance differences between the two modes of reasons,verification of the loop traversal mode can obtain better tracking accuracy,It is also explained that the improvement of tracking accuracy in the traversal mode is the conclusion at the expense of frequently adjusting the ranging sensor pointing.Based on the application requirements of the space-based infrared optical monitoring system under the main passive joint detection system,this paper puts forward the information processing algorithm applicable to the detection system,and the research results enrich the algorithm theory of surface target tracking,ballistic parameter estimation,stereo trajectory output,etc.It provides technical support for the design and development of information processing subsystems of space-based optical monitoring systems.