SAR image classification based on texture feature fusion

This paper presents a method for feature extraction and classification of synthetic aperture radar (SAR) images. This proposed method consists of three steps. First, two kinds of texture features are extracted for SAR image, which are the gray level co-occurrence matrix (GLCM) and Gabor filters (GFs). Second, these two kinds of extracted feature vectors from the first step were fused using the canonical correlation analysis (CCA) to reduce the dimensionality of the feature spaces. Third, the SAR images are classified with the support vector machine (SVM) in the fused feature space. The experimental results demonstrate that the proposed SAR classification method obtains good classification performance and the dimensionality reduction of CCA leads to high efficiency.