Classification Of Polarimetric Synthetic Aperture Radar Images

Synthetic Aperture Radar (SAR) image classification is a key technique of SAR image interpretation. Fast and accurate SAR image classification is a very pivotal part in the practical applications. In this dissertation, we research on some classification methods using polarimetric SAR (POLSAR) data.USA keeps ahead in the whole SAR image interpretation field in the world. Some other countries, such as Canada, German and Japan, are researching on SAR image interpretation. SAR information processing technique of China is ultimately underway.In the practical applications of SAR image classification, we have little prior knowledge about the classes and we rarely use polar information. Besides, many classification errors are caused by mixed pixels and speckle noise of the SAR image. Since the manual technique of selecting training sets could be difficult when ground truth is not available, we usually use unsupervised classification methods to classify the image automatically. In this dissertation, we study some unsupervised classification methods through three aspects, such as preprocessing, the selection of classifier and feature extracting. The effectiveness of these methods is demonstrated using SIR-C/X-SAR (Spaceborne Imaging Radar-C and X-band Synthetic Aperture Radar) data.In this dissertation, we study an unsupervised classification method based on fuzzy set theory. This method is a combination of the usage of polarimetric information of POLSAR data and the unsupervised classification method based on fuzzy set theory. We use image quantization and image enhancement techniques to preprocess the POLSAR data. We then use the polarimetric information and fuzzy c-means (FCM) clustering algorithm to classify the preprocessed images.Besides, we study an unsupervised classification method based on polar decomposition. This method is a combination of the polar decomposition and the...