Texture based classification of multidate SAR images ‐ a case study

Abstract

Texture based classification of multidate ERS‐1 SAR image was carried out. It was found that multidate data aided in enhancement of classification accuracy. The overall classification accuracy was 64%. When speckles were reduced, the accuracy obtained was 53.2%. Principal component image of multidate SAR image gave an accuracy comparable to that of speckle filtered image. Discrimination ability was also examined and it was found that though despeckled image gave slightly better separability measure, the classification accuracy came down by about 10% as compared to that for original image when the texture based classification technique was used, which can be explained because an unfiltered image contains more information about texture.