Application of region growing algorithm for classification of IRS multispectral image

The digital classification techniques produce images with noise and not so well defined interregional boundaries. In order to eliminate these defects and to enhance the quality of thematic maps post processing algorithms are generally applied. The IRS sensor LISS-data is used in this study and the information fusion method is applied to validate its utility for data classification. The two image products obtained after maximum likelihood classification and Sobel´s filter are merged using HIS transformation. The region growing algorithm available in literature is applied by following the conditions used in stopping the region-growing process