Detecção de mudanças na mancha urbana de Monte Carmelo - MG utilizando imagens multiespectrais

Abstract

Monte Carmelo is a Brazilian municipality of the state of Minas Gerais, which traditionally has economic activity linked to the production of tiles, bricks, ceramic artifacts and coffee production. In 2011, the advanced campus of the federal university of Uberlandia settled in the city, where today five baccalaureate courses are already taught. Due to these factors, there has been a growth in the municipal, which is already a reality for residents, but needs to be monitored and analyzed the needs of urban infrastructure, social facilities, services, among other components that make up the municipal directorial plan. In this sense, studying the intensity of the expansion is fundamental for the planning and articulation of public policies. Within this context, the use of orbital images with high spatial resolution is a viable and efficient alternative, being a strategic option that saves financial resources, since from them it is possible to estimate the growth of the urban spot precisely. Satellite images cover large areas with a regular temporal resolution, providing spatial and temporal information of several targets, widely used in change detection studies. Thus, the objective of this work is to analyze and quantify the growth of the Monte Carmelo urban network, using a time series of RapidEye images, from 2011 to 2018. For this purpose, pre-processing techniques of multispectral data were used, such as: correction geometric, atmospheric correction and radiometric calibration. After the influences of the errors were minimized and modeled the next step was to classify the images, and the objective was to find a way to highlight the urban stain of the other elements of the image, using two methods of classification: Segmentation of images and Artificial Neural Networks. In the image segmentation process, an equation was created, through histograms analysis, that can segment the urban part of the image.