Feature Selection Method Based on Mutual Information for Polarimetric Synthetic Aperture Radar (PolSAR) Image Classification

In recent years, Polarimetric Synthetic Aperture Radar (PolSAR) image classification has been cited as one of the most important applications of images classification. Therefore, in order to achieve the best result of PolSAR image classification in this article, a new feature selection method will be proposed based on mutual information theory. In the proposed method, the features that are extracted from PolSAR images will be used to obtain an initial class map. Then, each feature will be ranked based on mutual information. In the next step, the best features will be selected by using the proposed method accurately. The results that are obtained on the real PolSAR image of the Flevoland area prove an increase in the classification accuracy of the proposed method compared with other methods that are used in this research.