Research Projects

Extraction of important information from a satellite image or a set of satellite images (such as 2D and 3D building parameters and individual tree boundaries, number, height, canopy shape etc., water bodies, roads and their width, road types, tree plantations, parks, play grounds, built up area, in a habituated (urban or rural) area, in-land water bodies, information on cultural heritage and tourism sites, etc.) is important for various purposes (inventorisation, planning, management and decision making) at various levels. Availability of multi and hyper spectral (optical/microwave), multiresolution/sensor, multi-temporal from multiple platforms (space based aerial/ ground based/ UAV/ LiDAR) coupled with appropriate automated methods could fulfil the requirement of information extraction from satellite images. Several potential methods such as neural networks and deep learning, natural intelligence based methods, object based methods considering spatial spectral classification analysis, contextual methods, appropriate fusion (at pixel/ feature level) of multi-source data, need to be explored for effective utilization of the huge amount of multi-source data available from remote sensing satellites.