Sar Images Segmentation Based On Quantum Evolution Feature Selection Algorithm

SAR (Synthetic Aperture Radar) images segmentation plays a key role in the analysis for target detection, recognition and image compression. SAR images contain plenty of information such as edge features, shape features and texture features. This paper's work was mainly based on how to make use of these features.This paper firstly studied the theories of image segmentation and the commonly used methods of image segmentation in detail, discussed the different features in accordance with their classification, and discussed the advantages and disadvantages of various methods of image segmentation.Then SAR image segmentation was analyzed in the way of feature extraction. Texture is the most important information of images. This paper studied three methods of texture extraction, GLCM, the undecimated wavelet decomposition and texture extraction based Contourlet transform. According to the results of experiments which were done in the paper, the combination of GLCM, undecimated wavelet decomposition and texture extraction based Contourlet transform was proved to be excellent in SAR image segmentation.At last SAR image segmentation was analyzed in the way of feature selection in this paper. This paper studies one method of feature selection, which is based on Quantum Evolution Algorithm. This method can get the best feature subset by searching the feature set using Quantum Evolution Algorithm. This paper provided the results of experiments to prove the method is excellent in SAR image segmentation.