GlobVision has developed an operational software application for three main purposes: (1) Land cover mapping and change monitoring, (2) wetlands classification, and (3) monitoring water extent, using proprietary state-of-the-art multi-temporal SAR processing and machine learning techniques. The software uses RADARSAT-2 fully polarimetric SAR images and other ancillary sources of remote sensing data as input to perform land cover mapping and wetlands classification. The developed methodologies employ multi-source and multi-temporal image processing, feature extraction and classification techniques. First, multi-temporal and optical data are co-registered and then optimal subsets of multi-temporal features are identified for various areas of interest and combined to improve classification and mapping accuracy.
The use of GlobVision’s tool results in the improved accuracy of land cover mapping and wetlands classification compared to the most recent results by other research groups, as validated by Parks Canada. GlobVision’s approach automates all of the steps involved in land cover and water extent mapping and change monitoring – from reading images to image pre-processing, feature extraction, feature selection and dimension reduction to multi-temporal classification. Our software provides a unique and user-friendly capability for accurate and large-scale land cover mapping and change monitoring within an operational environment.
For more information, click here