RESUMO A extensão da malha rodoviária brasileira tanto em pistas simples quanto em pistas duplas ou alargamento de rodovias, tem aumentado significativamente junto ao crescimento populacional, e os impactos causados em função das construções e reformas vêm sendo constantemente discutidos sob o ponto de vista de suas consequências ambientais. Rodovias nunca deixarão de existir, tampouco novas construções, contudo, a implantação de medidas de controle e monitoramento ambiental podem reduzir os impactos negativos evitando danos irreversíveis ao meio ambiente. Este trabalho apresenta o desenvolvimento de uma metodologia para classificação de regiões em imagens do entorno de rodovias com base em Processamento Digital de Imagens (PDI) e Lógica Fuzzy a partir do uso dos descritores de cor e textura que melhor caracterizam os variados tipos de cobertura do solo. Para isso foram extraídas 600 amostras (recortes de imagens) de dimensões 29×29 pixels do entorno da rodovia Raposo Tavares, sendo 100 amostras para cada grupo de cobertura do solo considerados neste trabalho: residências, indústrias, rodovias, solo exposto, vegetação rasteira (gramíneas) e vegetação arbórea (matas). A partir dessas amostras foi construído um SIF (Sistema de Inferência Fuzzy) para classificação dos tipos de cobertura do solo. Ao se aplicar este sistema nas 600 amostras obteve-se uma matriz de confusão dos resultados e um índice kappa igual a 0,9197, que mostra a eficiência da metodologia desenvolvida.

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.

Computed Tomography (CT) has made possible visualization of physical structures in the interior of an object in their real relative spatial position and orientation. Another important contribution of CT is to greatly improve abilities to distinguish regions with different gamma ray transmittance and to separate over-lying structures. The mathematical problem of CT imaging is that of estimating an image from its projections. In this work, the experimental setup was performed using the Mini Computerized Tomograph of Uniso (MTCU). This tomograph system operates with a gamma ray source of 241Am (photons of 60 KeV and 100 mCi of intensity) and a NaI(Tl) solid state detector. The system features translation and rotation scanning modes, a 100 mm effective field of view, 1 mm of spatial resolution and 5 % to 10 % of density resolution. The image reconstruction problem can be solved using two different algorithms: Algebraic Reconstruction Techniques (ART) or Discrete Filtered Backprojection (FBP). The three-dimensional image reconstruction method discussed in this work, involves obtaining two-dimensional (2D) gamma ray tomography images and then combining these images into a three-dimensional (3D) volume data. An opacity, shading and color is attributed to each volume element (voxel) and the resulting scene is projected in a picture plane to be display in a monitor. Our implementation was developed to be used with MTCU data and was based in the Ray casting volume rendering technique. The use of such technique for nondestructive evaluation is a powerful tool to enable a visual trip inside an object without physically opening or cutting it. Experimental and theoretical methods used are discussed and results of experiments using the 3D reconstruction techniques are presented. A particular use of this technique to study concrete properties, such as stones distributions, visualization of structural occurrences inside concrete samples is discussed.

Computerized Tomography (CT) refers to the cross sectional imaging of an object from both transmission or reflection data collected by illuminating the object from many different directions. The most important contribution of CT is to greatly improve abilities to distinguish regions with different gamma ray transmittance and to separate over-lying structures. The mathematical problem of the CT imaging is that of estimating an image from its projections. These projections can represent, for example, the linear attenuation coefficient of gamma-rays along the path of the ray. In this work we will present some new results obtained by using tomographic techniques to analyze column samples of concrete to check the distribution of various materials and structural problems. These concrete samples were made using different proportions of stone, sand and cement. Another set of samples with different proportions of sand and cement were also used to verify the outcome from the CT analysis and the differences between them. Those samples were prepared at the Material Laboratory of Faculdade de Engenharia de Sorocaba, following the same procedures used in real case of concrete tests. The projections used in this work was obtained by Mini Computerized Tomograph of Uniso (MTCU), located at the Experimental Nuclear Physics Laboratory at University of Sorocaba. This tomograph operates with a gamma ray source of 241Am (photons of 60 keV and 100 mCi of intensity) and a NaI(Tl) solid state detector. The system features translation and rotation scanning modes, a 100 mm effective field of view, and 1 mm spatial resolution. The image reconstruction problem is solved using Discrete Filtered Backprojection (FBP).