Detection And Tracking Of Weak Moving Targets For Space Based Warning Radar

Warning radars are widely used in military and civil fields.It is necessary for space-based warning radars to surveil targets on the land or in the ocean,air or space.Particularly,the trend estimation for threated targets is extremely important.However,in the environment of low signal to noise ratio(SNR),it is difficult to detect,track and locate the maneuvering targets.Many researchers all over the world are interested in these problems.Increasing integration time can improve SNR.However,for the high-speed motion of weak targets induces range migration in the long-time period,the energy of signal may not be accumulated effectively.Track before detect(TBD)can make most use of information,and in the environment of low SNR,it is especially applicable for TBD to detect and track targets.In addition,it is required to set up new model and use improved algorithms when detection and tracking the extended targets.This thesis investigates the challenging problem of detection and tracking for the weak targets in the terms of mode selection of space-borne radar tracking target,multi-target track detection using image processing in the environment of low SNR,a new detection-tracking integration method based on finite frame memory and iterative updating and extended target detection and tracking in wide area surveillance(WAS).The main content of this study can be summarized as follows: 1.On the basis of orbital motion knowledge,to address the problem of tracking mode selection by using space-borne radar,the coordinate systems are established first.Secondly two modes are designed,one is first converted then tracking the absolute state of motion,the other is first tracking the relative state of motion then coordinate conversion.Then the analysis of constraints for the two tracking mode is discussed,including the characteristic of measuring error transferring and the difficult for establishment of state modeling.Finally,a simulation experiment is made for the comparison of tracking accuracy of the two designed modes.The simulation results show that,compared with first tracking the relative state of motion in measurement coordinate system then coordinate conversion,first converted then tracking the absolute state of motion in inertial coordinate system can obtain significant improvement in tracking performance.The research can provide the design basis for space-borne radar tracking mode selection.2.A new method for multi-target track detection is presented.The first stage is pre-processing,i.e.,a range-time image will be accessed.Then,edge detection based on phase congruency model is performed.Secondly,in order to obtain consistent orientation of the edge connected region,adaptive region growing is carried out.Finally,combining clearing operation and least squares fitting to detect multiple targets and estimate the target parameters.The method has the advantage that it can detect multi-target track under low signal noise ratio(4dB).The effectiveness of the algorithm is confirmed by extensive simulation.3.Two different approaches based on finite frame memory and iterative updating are addressed for dim moving target track detection.One method is a new detection-tracking integration method combination finite frame memory and iterative updating.Firstly,aiming at finite frame memory N-frame original data,adaptive image domain track initiation is done,and the targets' points set can be gotten.According to the state model state prediction is processed.Combined with the detection results of the current window and the prediction value,the size and location of the next detection window for each target is determined.Then,track management of the overlapped part is put up for the purpose of track association corresponding to the same target.The tracking precision can be increased,and then the detection tracking integration processing are completed.Another one is an adaptive target tracking method based on parameterized trajectory enhancement,in which the image processing and the curve fitting are combined.First of all,nonparametric track initialization is processed based on multi-frame data to gain the initial trajectory and parameters of the target.Then the edge enhancement masks are designed by using the parameters of the target's state.Afterward,the designed masks are used to operate the parameterized trajectory enhancement.Finally,the threshold judgment is used to decide whether the moving target exists,therefore the results of tracking can be obtained.The algorithm possesses the characteristics of close loop feedback between the state of target's trajectory and the track enhancement masks,which can improve the accuracy of detection and tracking and realize joint detection and tracking in low signal to noise ratio environment.The performance of the proposed algorithm for detection and tracking dim targets is improved,which can be seen from the simulation results.4.When wide area surveillance(WAS)by warning radar,first of all,in order to select training samples,combining the magnitude histogram and interferometric phase histogram with region segmentation.Following,the adaptive matched filtering(AMF)is used to suppress the ambiguous clutter.The geomorphic feature of images is utilized,and then a registration algorithm combined SIFT with region segmentation is proposed.Multi-frame of different perspectives images are registered.On the one hand,with the development of radar technology,the target size is larger than the radar resolution;on the other hand,the problems of migration through resolution cell and azimuth defocusing as a result of mismatching moving parameters.It is necessary to consider the challenge of extended target detection and tracking.To address the problem of mismatching moving parameters extended target,make use of the relationship between target shape parameters and motion parameters to classify and identify the multi-targets.Simulation results show that the proposed algorithm can track multiple extended targets effectively.