This seminar work looks at the theoretical basis for automatic knowledge-based preliminary classification of multispectral remotely sensed satellite imagery using the Satellite Image Automatic Mapper™ (SIAM™) developed by Dr. Andrea Baraldi. SIAM™ was used to pre-classify one Landsat-8 image and one Sentinel-2 image of Salzburg at all available semantic granularites (18/48/96 semi-concepts). The paper can be read here in its original form, not including post-submission edits after further discussions with Dr. Andrea Baraldi.
The theoretical background and concepts of preliminary classification of multispectral remotely sensed satellite imagery are reviewed based on the Satellite Image Automatic Mapper (SIAM™), developed primarily by Dr. Andrea Baraldi. Landsat-8 and Sentinel-2 images from August 2015 of the city of Salzburg, Austria were pre-classified following radiometric calibration into top of atmosphere (TOA) reflectance. The operational, automatic, prior knowledge-based, multi-sensor, multi-resolution, near real-time SIAM™ was used to pre-classify the image pixels into semi-concepts at three semantic levels. Results from both Landsat-8 and Sentinel-2 are compared with each other between semantic granularities. Vegetation binary masks were also created using SIAM™, which may have use particular relevance to automated land use change analysis. A few applications for this type of image pre-classification are briefly explored.
Baraldi, A. (2011) Satellite Image Automatic Mapper™ (SIAM™). A turnkey software button for automatic near-real-time multi-sensor multi-resolution spectral rule-based preliminary classification of spaceborne multi-spectral images. Recent Patents on Space Technology, 2011(1), pp. 81-106.
Baraldi, A., and Boschetti, L. (2012) Operational automatic remote sensing image understanding systems: beyond Geographic Object-Based and Object-Oriented Image Analysis (GEOBIA/GEOOIA). Part 1: Introduction. Remote Sensing, 4, pp. 2694-2735.