Análise Automática do Uso do Solo no Entorno de Rodovias Usando uma Abordagem Fuzzy

ABSTRACT

The extension of the Brazilian road network, both in simple and multiple lanes and in widening of highways, has increased along with population growth, and the impacts caused by important constructions and reforms are constantly discussed from the point of view of their consequences environmental issues. Highways will never cease to exist, nor will new construction. However, the implementation of control and environmental monitoring measures can reduce negative impacts by avoiding irreversible damage to the environment. This work, aims to develop a methodology for classifying and replacing images of highways around based on Digital Image Processing and Fuzzy Logic to extract color and texture descriptors from various types of soil cover. For this, 600 (29×29 pixels) dimensions (image clippings) were collected from the surroundings of the Raposo Tavares highway, 100 for each ground cover group, forming the basis of the thesis study: residences, industries, highways, exposed soil, undergrowth (grasses) and tree vegetation (forests). From these samples, a FIS (Fuzzy Inference System) was built to classify the types of soil cover. When applying this system to the 600 samples, a confusing matrix was obtained and a calculated kappa index equal to 0.9197, which shows the efficiency of the developed methodology.

Keywords:
highways; color and texture; fuzzy; Mamdani; classification and segmentation