Progressive Morphological Filtering for Land Cover Clusttering Based on SAR Imagery

This paper presents a progressive morphological filtering (PMF) approach on land cover clustering based on Synthetic Aperture Radar (SAR) remote sensing image. For SAR image containing only intensity and texture information, pixel-based classification methods produce less satisfactory results due to speckle noise in SAR image. While morphological filtering operation can be used to improve such result. In the classification step of the SAR image, K-means clustering method is used in combination with Decision-Tree method for classification of the clusters. The proposed PMF approach can be used within the classification steps to not only reduce classification errors caused by speckle noise, but also improve the clustering visualization effects with flexibility in meeting various purposes, such as land cover overview for land surveying or planning. Two SAR images of Singapore are used to validate the performance of the approach.