An approach for vegetation mapping and pixel-based change detection for IRS-1C LISS III data

Monitoring and tracking vegetation changes in vast and remote areas is a difficult task. Accurate extraction of existing vegetation is the primary step for better statistical assessment. An approach for vegetation mapping is proposed which automatically explores spectral characteristics of connected components in an image. Different frames of multi-spectral Indian Remote sensing IRS-1C LISS III (Linear Imaging Self-Scanning Sensor) data of different times were used for comprehensive view of vegetation cover. First, quantum of vegetation cover was extracted using band indexing and brightness manipulating techniques for geometrically registered multi-temporal image pair. Secondly, pixel based image differencing method was employed to determine the temporal changes. The accuracy of vegetation extraction was found to be satisfactory by comparing the results with Normalized Difference Vegetation Index vegetation extraction method. Pixel based change analysis technique was assessed statistically by comparison with other change detection techniques like image differencing and band ratioing. Experimental resultant images have demonstrated the efficiency for the proposed approach for the segmentation of vegetation. A statistical value of pixel based change analysis was found to be satisfactory for medium resolution images like IRS-1C LISS III imagery.