Study On ISAR/InISAR Motion Compensation And Imaging Algorithms

With the improvement of application requirements and the military defense capability,the radar imaging technique has played a crucial role due to its distinctive capabilities of high-resolution imaging at long range,day and night,and in all weather conditions.Because of its special ability of information acquisition in remote sensing for non-cooperative targets,inverse synthetic aperture radar(ISAR)has become one of the most important means of our country for surveillance and strike in airspace,aerospace and ocean field.Follow the development tendency of wider-swath and higher-resolution imaging in ISAR imaging technique,and combine the often innovative radar system,the working patterns and data acquisition modes are gradually developed to multiple functions,multiple dimensions and collaborative networks from ones,which greatly challenge the current ISAR imaging system in high-resolution image formation and target parameter extraction.Under the support of National 973 Plan project(Research on theory,system and method of Sparse Microwave Imaging)and lateral project(Three-dimensional geometry of space targets for interferometric ISAR),this dissertation focuses on the problems of motion compensation,parameters estimation for maneuvering targets and three-dimension imaging.The primary goal of this study is to enhance the accuracy of automatic target recognition.The outline of this dissertation can be listed as follows:(1)Approach to joint translational and rotational phase auto-focusingTo deal with the hypothesis that the translational motion compensation has already been successfully realised,the first part studies the robust autofocusing approaches of ISAR imaging by taking the residual translational phase errors into account.For this purpose,the translational and rotational phased errors are modelled as the range-invariant and range-variant forms based on the interaction between them.This scheme uses a minimum entropy criterion to construct the cost function from a statistical point view.Next,a coordinate descend method is employed to solve the optimisation for joint phase error correction,which is executed using a quasi-Newton method in each iteration.It is important to note that the conventional methods should be employed to realise a coarse stimation of translational phase errors for efficiency of optimisation solution.Comparing with the conventional approaches without joint phase error corrections,the proposed algorithm promises a better performance of ISAR imaging in a much more efficient and effective way.(2)Parameter estimation and high-resolution image for non-uniformly rotated targetsIn the second part,the ISAR imaging approach for uniformly rotated targets is extended to the case of non-uniformly rotated targets.Here,taking the linear frequency modulated model as an example,we present a joint rotational parameter estimation and high-resolution imaging algorithm via Matched Fourier Transform(MFT)in Chapter 4.The essence of this approach is to replace the linear transform of FFT with the parametric transform of MFT.Herein,by exploiting the strong sparsity of ISAR image in MFT domain,a sparse measurement matrix is constructed to recursively compensate the range and azimuth coupled phase error through sparse signal processing scheme,and the high-resolution image and rotating parameters are alternately solved eventually.Meanwhile,the probability of target recognition can also be improved by accuracy size parameters of targets.Then,the image in range-Doppler domain can be transformed into range-azimuth domain via azimuth scaling factor which derivated from the estimated rotated parameters.(3)Joint sparsity constraint In ISAR imaging for 3-D geometry of maneuvering targetsIn the third part,the single-channel 2-D high-resolution ISAR imaging approach described in Chapter 3 and 4 is extended to the case of joint multi-channel one.In this context,the interferometric processing method,which is realized via three antennas in a special coordinate configuration,is employed to achieve the distinguishable height dimension using the interferometric phase of ISAR images acquired from different phase centers.Beyond the limits of information acquisition loss,which produced by the projection of targets in the image plane,the 3-D geometry reconstruction method can provide the more reliable information for target identification.To ensure a high coherence between multi-channel images,it is necessary to calibrate the channel mismatch by combining the three channel data under a joint Chirp-Fourier dictionary.In the scheme,joint the multi-channel In ISAR 2-D image formation is treated as a joint-sparsity constraint optimization.Then,the optimization is solved using a modified OMP algorithm to realize 2-D imaging,parameter estimation for linear frequency modulated form and scattering center extraction.Finally,joint the estimation of 3-D geometry and rotation motion is performed in an iterative manner to effectively remove the outliers and estimation errors.(4)The method of effective rotation vector estimation and 3-D imaging via In ISARFor the maneuverability of non-cooperative targets and energy losses exist in the traditional 2-D ISAR image,the algorithm of effective rotation vector estimation and 3-D imaging is proposed in Chapter 6.Thus,a real sense of 3-D ISAR imaging is achieved,and which provides basis for the target classification and identigication.At last,the numerical results validate the effectiveness of the proposed method.