Sar Image Feature Data Extraction And Sar Image Segmentation

Synthetic aperture radar （SAR）image information more and more, how to makebetter use of synthetic aperture radar images provided information and new technologytheory, to realize the synthetic aperture radar image feature extraction and accurateprecise interpretation, is a synthetic aperture radar image in military target tracking andidentifying the bottleneck problem.Synthetic aperture radar image has many different from ordinary optical imagefeatures: for example, large dynamic range, serious speckle noise and rich textureinformation.Due to the fractal characteristics of synthetic aperture radar image processing isdescribed having a texture roughness, resistant to speckle noise ability and the humanvisual perception consistent characteristics, so the fractal theory on its own uniquedescription image mode, for synthetic aperture radar image processing has opened up anew way.Fractal characteristics of fractal dimension, is the description of image fractalfeature is very useful tool, based on the differential box counting dimension hollow boxeffect, we use probability computing method based on fractal Brown model the truedifferential box counting algorithm, to the greatest extent, reduce the empty box onfractal dimension calculation of the impact.Gray level co-occurrence matrix is through the research of gray and spatialcorrelation characteristics to describe the texture of a commonly used method, but alsohow we make good use of texture information in synthetic aperture radar imageprocessing on the important aspects.Based on an amount of experimental analysis of the effect of synthetic apertureradar image gray level co-occurrence matrix (GLCM) distance, window size and otherfactors, establishes the extraction of synthetic aperture radar image gray-levelco-occurrence matrix element size, and aimed at the big window to extract a smallerresolution synthetic aperture radar image by image information loss in thedisadvantages, design a using dynamic sliding window to extract gray-level co-occurrence matrix, effectively reduces the information loss in the faults occur.According to the optical image segmentation method in the processing of syntheticaperture radar images usually can not get a good result.Extraction of synthetic apertureradar image texture characteristics and fractal characteristics, this paper summarizes thesynthetic aperture radar image segmentation algorithm, realized based on gray-levelco-occurrence matrix fuzzy C mean synthetic aperture radar image segmentationalgorithm.The algorithm combines the gray-level co-occurrence matrix and syntheticaperture radar image gray value syndrome, using fuzzy C means clustering (Fuzzycluster method, FCM) algorithm for eigenvalue cluster based on synthetic aperture radarimage segmentation.At the end of the thesis were completed using the fractal dimensionof the synthetic aperture radar SAR image segmentation, the experiment proved theeffectiveness of the two methods.