FIS Toolbox – A Geoprocessing framework for predicting climate change with spatio-temporal database
Fuzzy Inference Engine is a framework that develops a rule base for performing spatial segmentation of objects in Satellite Imagery, Landsat-8. The engine is implemented using concepts and principles of Fuzzy Inference and Hierarchical Scale Space with a detailed set of computer vision algorithms in java, for the delineating of objects in a given scene. This geoprocessing toolbox is referred to as Fuzzy Inference System (FIS). Geo-processing and Geo-computation are two major research areas in computer vision and pattern recognition that are driving the concepts in High-Performance Satellite Imagery processing and machine learning. In the geospatial domain there are not detailed research methodologies developed for Landsat-8 Thematic classification using state-of-art Digital Image Analysis techniques. In this research endeavor, we propose and implement a Geo-computation toolbox for edge boundary detection, feature sampling and Laplacian of Gaussian Smoothing, Hierarchical Scale Space Generalization with higher level Specialization for multi-scale analysis, Image Compression with Pyramid Tiling, Scene objects delineation with Region contouring and labelling, and finally, computation within R Statistical Software. Multivariate image statistics are derived for understanding of the inherent geometrical and pattern recognition feature objects. Specifically, Spatial Autocorrelation, spatial feature variation using variograms and change detection visualization in vegetation with Normalized Difference Vegetation Index. Initial results are encouraging with an intent of developing subsequent version of the toolbox using data science fundamentals.