Segmentation of Built up area from SPOT 5 multispectral satellite images

Segmentation of Built up area from a satellite image is an initial step for a variety of applications. In this paper we propose a novel approach for segmenting built up area mask from SPOT 5 multi spectral satellite images with 2.5 meter resolution. The proposed algorithm uses information theoretic and multi resolution techniques for segmentation. The results of proposed algorithm demonstrated it's effectiveness for the segmentation of built up area from large sized satellite images. The validation results for segmentation using sensitivity, positive predictive value and dice coefficient show a significant improvement over supervised classification results of a commercial software. Also it was observed that the wavelet transform produced better results than the scale space technique when used in the proposed algorithm for multi resolution decomposition.