Advanced classification of remote sensing high resolution imagery. An application for the management of natural resources

Abstract

In the last decades, there has been a decline in ecosystems natural resources. The objective of the study is to develop advanced image processing techniques applied to high resolution remote sensing imagery for the ecosystem conservation. The study area is focused in three ecosystems from The Canary Islands, Teide National Park, Maspalomas Natural Reserve and Corralejo and Islote de Lobos Natural Park. Different pre-processing steps have been applied in order to acquire high quality imagery. After an extensive analysis and evaluation of pansharpening techniques, Weighted Wavelet 'à trous’ through Fractal Dimension Maps, in Teide and Maspalomas scenes, and Fast Intensity Hue Saturation, in Corralejo scene, are used, then, a RPC (Rational Polymodal Coefficients) model performs the orthorectification and finally, the atmospheric correction is carried out by the 6S algorithm. The final step is to generate marine and terrestrial thematic products using advanced classification techniques for the management of natural resources. Accurate thematic maps have already been obtained in Teide National Park. A comparative study of both pixel-based and object-based (OBIA) approaches was carried out, obtaining the most accurate thematic maps in both of them using Support Vector Machine classifier.