Using Remote Data Mining on LIDAR and Imagery Fusion Data to Develop Land Cover Maps

Remote sensing based on imagery has traditionally been the main tool used to extract land uses and land cover (LULC) maps. However, more powerful tools are needed in order to fulfill organizations requirements. Thus, this work explores the joint use of orthophotography and LIDAR with the application of intelligent techniques for rapid and efficient LULC map generation. In particular, five types of LULC have been studied for a northern area in Spain, extracting 63 features. Subsequently, a comparison of two well-known supervised learning algorithms is performed, showing that C4.5 substantially outperforms a classical remote sensing classifier (PCA combined with Naive Bayes). This fact has also been tested by means of the non-parametric Wilcoxon statistical test. Finally, the C4.5 is applied to construct a model which, with a resolution of 1 m, obtained precisions between 81% and 93%.