Reconstruction Method Research On High-Resolution SAR Scene Distribution

SAR (Synthetic aperture radar) system has the advantage of all-weather, all-day,large covering area and the ability target imaging and recognition. Compared with thelow-resolution SAR, the high-resolution SAR can well show fine structures of radartarget, which is reflected the differences in texture, so it has a wide applicationprospects in the modern military war. The research about high-resolution SAR scenedistribution is still in flux, so it has important theoretical and application value.First of all, based on the analysis of the high-resolution SAR imaging technology,the target’s RCS characteristics of narrow-band and wide-band radar are analyzed.The high-resolution SAR features, including speckle and target strong scattering center,and its the uncertainty presentation are selective discussed.Secondly, taking account of high-resolution SAR speckle and RCS, the SARstatistical distribution model is studied. According to real SAR scene’s verification test,the model’s usability is analyzed. The relevant statistics random sequence generationthough a ZMNL (zero memory nonlinearity) is raised. Considering the statisticaldistribution in different resolution SAR image, so the novel SAR scene reconstructionmethod based on pyramid is proposed, which analysis the fine structure of SAR texture.Finally, the SAR scattering center model and scattering centers reconstructionmethod are studied. In view of the methods existing scattering point’s outliers and thebig errors in the point’s location similarly, the modified MEMP (matrix enhancementand matrix pencil) algorithm is put forward. When existing same location in anydirection, the improved algorithm uses the block matrix decomposition to revise theparameters and eliminate the false scattering point to achieve the goal that the targetscattering center reconstruction image is similar with the original target. Simulationresults show that the proposed algorithm can reconstruct the target geometry structureeffectively and the estimation precision is higher than MEMP algorithm.