Vol. 9 No 6 December 2001
The software allows labelling multispectral satellite data in terms of corresponding ground materials and display of a coloured multispectral image and different colours overlaid on multispectral image to represent different classes in which the image is classified. The input to this package is full IRS-1C/1 D LISS-III multispectral image. The software allows selection of training samples of different ground materials and knowledge base of the same depending upon the application, which can be updated. The software also provides a facility to extract the samples from a zoom window when the sample size considered to extract the samples is same. Modules for error analysis and refinement of training samples have also been provided for error free training data which in turn improves classification accuracy. The output of this software package is a material labelled image which itself is a product.