Mapping of Hydrographic Networks from Multispectral Imagery using Neural Networks and Principal Curves

This paper presents two techniques developed in efforts to fully automate the generation of hydrographic maps from remotely sensed imagery. The methods presented here consist of two techniques using self-organizing networks. These experimental techniques have been explored for their application in automated generation of graphical representations of hydrological objects. The first technique involves object extraction from multi-spectral satellite imagery, while the second is required for the automatic mapping of the extracted water basins. These methods effectively manage occlusions and data discontinuities. The experimental test results using Landsat-7 ETM+ images demonstrate the accuracy of the proposed approach and its potential for fully automated mapping of hydrological objects