Geospatial Raster Data Processing Applying Neural Networks
摘要
This work proposes applying machine learning techniques to classify raster geospatial data, which have traditionally been classified using classic segmentation and clustering techniques. In the first instance, we propose a neural network model using a multilayer perceptron model to carry out the classification task. Such a model can be trained directly by the data samples due to the nature of the raster geospatial data, which has several bands that serve as a spectral signature and will be used for expressed training tasks.