Object Based Image Analysis For Delineation Of Slope Units

This research focuses on proposing an approach for delineation of slope units from topographic parameters of terrain slope, slope aspect and topographic position index (TPI) extracted from ALOS World 3D digital elevation model (DEM). The research was carried out through segmentation processes at four different levels with varying parameters values of scale, size and compactness on RGB composite (TPI, slope, aspect). D-matrix was used to find the optimal parametrization for accuracy measures and showed that the lowest value of $D$ lies with the scale parameter as 50, size as 0.3 and compactness as 0.7. The segmentation results obtained with these parameter values were classified into two landform classes of flat terrain and slope units and compared with the reference watershed boundary delineated using soil and water assessment tool (SWAT). The result showed higher agreement in terms of accuracy with changing slope positions of ridge line, drainage line and valley that can be further used for enhanced landslide susceptibility maps.