Automatic extraction and analysis of lineament features using ASTER and Sentinel 1 SAR data

In recent years, automatic lineament extraction has gained huge popularity as it circumvents the issues such as errors and time constraints associated with manual extraction procedure. Another impediment in the lineament extraction process is that linear features existing near sub-surface or occluded due to vegetation cover cannot be identified solely by optical data. To mitigate these issues, a multi-sensor fusion-based automatic extraction process using EO-based optical and SAR data is proposed. Use of moderate resolution optical data such as ASTER minimizes erroneous identification and misclassification of anthropogenic features as geological features. In addition to this, a component loading-based band selection approach is proposed to select the most informative bands from the available optical bands for improved extraction of lineament features. Finally, a new HSV-based mapping method for visualization and directional analysis of the extracted linear features is discussed. The results are validated using the published seismo-tectonic map of the region and field investigation studies which corroborates that the proposed algorithm can successfully extract most of the geological lineaments in the study area with minimum errors.