GLCM法による合成開口レーダ画像のテキスチャ解析の一方法

Two step GLCM法の提案

抄録

The SAR (Synthetic Aperture Radar), one of the microwave imaging radars, can offer a very high resolution image. It works effectively all day long in any weather condition without blocking by clouds. However, the conventional knowledge with visible light sensors can not be applied to the interpretation of SAR image, because of the property difference between two images.
Our final purpose is to find an interpretation technique of undefined objects on the SAR images. We have examined one method to extract texture features from SAR images so far. Improving GLCM (Gray Level Co-occurrence Matrix) method, we propose Two Step GLCM method. In this new method, the GLCM method is applied double to one image with different conditions upon the partial images.
This analysis yields an improved image features in comparison with the conventional simple GLCM method. Both micro-and macro-scopic image features are possible to be obtained separately. Further, the higher order features can be found on the image also.
The authors had a confirmation that Two Step GLCM method is effective for complex texture pattern analysis using model patterns and SAR images.