Semi-automated impervious feature extraction using built-up indices developed from space-borne optical and SAR remotely sensed sensors

In order to cope up with the emerging demand of interpreting and detecting urban features from remotely sensed imagery, this study proposes an innovative unique yet practically easy approach to extract impervious surfaces and features using certain built-up indices (BUIs) that are developed from space-borne optical and SAR remotely sensed sensors. Resourcesat-2 LISS III and HH/HV dual polarized L-band ALOS PALSAR data are respectively the space-borne optical and synthetic aperture radar (SAR) data used for designing the BUIs used for impervious feature detection. BUI generated from optical sensor uses green (G), red (R), near infrared (NIR) and short wave infrared (SWIR) channels simultaneously. BUI designed from SAR sensor uses horizontal-horizontal (HH) co-polarized and horizontal-vertical (HV) cross-polarized L-band backscatter coefficient/sigma nought (sigma degrees) data. The feature extraction performed by two new indices, one derived from existing indices using spectral-based information from optical sensors- an Impervious Built-up Index (IBUIopt); and the other from SAR HH/HV sigma degrees backscatter information- an Impervious Built-up Index (IBUIsar); are cross-checked and validated with Ground Truth (GT) data that revealed the performance of the indices designed and proposed in the study for three sample sites bearing completely different biophysical environments; for easy and rapid extraction of impervious land features in satellite imageries. Overall accuracies of impervious feature extraction using IBUIopt are 92.3%, 95% and 75.6% in comparison to 90%, 98% and 87.6% using IBUIsar for the three study areas respectively. The approach adopted in this study is henceforth acceptable due to its high accuracy, simplicity and reliability; and hence easy to replicate. (C) 2020 COSPAR. Published by Elsevier Ltd. All rights reserved.