Ice-Water Interpretation Based On ScanSAR Sea Ice Image

Sea ice interpretation is an important part of sea ice monitoring and forecasting, SAR (Synthetic Aperture Radar) is an important tool for sea ice monitoring and forecasting. To obtain land more information in a wider range in one scanning process, many SAR systems currently have the ability to work on the ScanSAR (Scanning Synthetic Aperture Radar) mode. In this mode, the radar system scanning width can reach more than 100 km, there is a big variation of incidence angle in the sea ice image acquired by this mode. In order to reduce the impact of variation of incidence angle for sea ice interpretation, it is necessary to perform a correction of the incidence angle effect before the interpretation of ScanSAR sea ice images, and then perform interpretation on the corrected image.Based on the correction method of incidence angle studied in the previous literatures, a new class-based correction is proposed in this paper. Sea ice SAR dataset testing in this study is collected from the Bohai Sea by ENVISAT ASAR To achieve a class-based correction, we firstly dividing the original image into several blocks on the range direction and performing a primary unsupervised classification which is based on block. The mask image of each sea ice type is then constructed using the classified image, and the proposed class-based locally linear mapping (LLM) technique is performed on the specific sea ice types. As the distribution of backscatter values in azimuth bands is discrete and non-linear, the LLM technique has to numerically sort the values of the specific target class in the azimuth bands, and approximately divide the sorted values into N subsets averagely to make the distribution locally linear. The locally linear mapping is therefore can be completed between the reference band and the azimuth bands. The results demonstrate that LLM-corrected ScanSAR images appear to have more detailed textures, and the natural signal variability in the radar data is preserved, which indicates that the LLM produces better results compared with the histogram-based-alike (HIST-alike) technique when correcting the incidence angle in the sea ice SAR data. In addition, this paper has discussed the method of determining the reference band and the correlation between the azimuth band width and the average rate of change of backscatter values.In this paper, we proposed MRF-SVM classification system by introduced the spatial context information based on Markov random field (MRF) into support vector machine (SVM) classifier. First, the system will regional SAR sea ice images; then we determine the strength of the edge using dual-threshold criteria; final, for the area of fuzzy edge, we improve model of spatial context. By correcting the bias of original problem of SVM, thereby we optimize the optimal hyperplane. The results demonstrate that the method can effectively improve the accuracy of ice-water interpretation based on ScanSAR sea ice image.