Classification of Vegetable Coverage Through Chromatic and Technical Characteristics of Artificial Intelligence Using Satellite Images

Abstract

Nowadays the use of information and communications technology is becoming a transversal tool, in the development of the different disciplines of science, the disciplines that are on the engineering side are not alien to this influence, a of these tools provided by information and communications technologies, are the techniques of artificial intelligence, which among its different uses and applications, is the ability to classify objects based on the analysis corresponds to certain patterns, which is why in the This work uses the neural networks to classify areas that correspond to the vegetation cover and areas that are deforested, under the analysis of the chromatic characteristics of the texture of the image, obtained through the earth observation satellites , these satellite images record the surface of the earth and represent them in an imag in multispectral, according to the characteristics of each spectral band, the changes that occur in the land cover can be distinguished, in this paper a study is carried out on the vegetation cover that is being affected by illegal mining in the jungle Peruvian, so that a registration and search of satellite images corresponding to the affected areas was carried out, with this an image bank was constructed to be able to analyze them and obtain the chromatic characteristics of uniformity, entropy and energy, with these it was possible to design a Neural network with three inputs and one output, to be able to classify between areas that have vegetation cover and areas that do not have vegetation coverage, the use of satellite images of the Lansat 8 mission was used using the Red, Green, and blue spectral bands in the combination of colors of natural color, the results obtained determine that you can work with the techniques of artificial intelligence in the recognition of the areas of the land cover, for our particular case such as the analysis of vegetation areas, Landsat 8 images were used because they had a greater coverage in terms of their spatial resolution so that a pixel in the Image corresponds to 30 meters on land, finally we present a graphical interface to analyze the areas and perform the classification.