Radiometric Correction Method On Area Effect Induced By Terrain For Spaceborne SAR Image

Compared with the traditional optical remote sensing image, SAR image canprovide a lot of target scattering characteristic, and has unique advantages in targetdetection and quantitative parameter inversion.However, radiation distortion causedby SAR imaging geometry and topographic relief will introduce errors for thebackscattering coefficient of ground objects. To study the quantitative relationbetween the backscattering coefficient and object tyoe, it is necessary to eliminate theinfluence of system and non-system factors on backscattering coefficient, and thetopographic relief is one of the factors that needed to be eliminated.Radiometric distortion caused by topographic relief is generally divided into twoaspects: one is the effect of area change. Topographic relief changed the scatteringarea of the same pixel, and then the same pixel received the different backscatteringpower. Another is the effect of local incidence angle change. The local incidenceangle could influence backscattering mechanism.For the first palce and with supportand guidance of the national863theme program called as object-orientedhigh-credible SAR processing system, this paper proposes some measures.The mainresearch and innovation contents are as follows:(1) Four terrain radiometric correction methods for SAR image were studied indetail, and the advantages and disadvantages of each method were summarized. ThenPALSAR data in DiQing experiment area and COSMO data in ChaYu experimentarea were selected to compare and evaluate the four terrain radiometric correctionmethods quantitatively and qualitatively. The results show that area integrationmethod is the most effectively terrain radiometric correction method.(2) This paper deduced the strict area integration formula in terms of themechanism of SAR imaging and proposed a complete terrian radiometric correctiontechnological procedure of SAR images based on area integration method andimaging automatic matching. The PALSAR data were used to detailedly describe theprocedure of area integration method in Diqing test area.(3) Although area integration could get the better correction results, it alsoappears the "garland" in front-slope area of corrected image. Although the ascendingand descending SAR image fusion could eliminate the absence of information inlayover and shadow regions, it also caused the uneven illumination problem in fusionimage. In order to solve the aboved two problems, a method of ascending and descending SAR image terrain radiometric correction is presented. Ascending anddescending quad-polarization Radarsat-2data in Lingshui test area were processedwith the proposed method. The corrected image and ascending and descending fusioncorrected image were classified using complex wishart unsupervised classificationmethod, and the classification results show that the new method significantlyimproves the SAR images classification accuracy.