Airton Marco Polidorio

Automatic shadow segmentation in aerial color images

A robust and efficient technique to segment shaded areas in aerial color images is presented here. This technique is based on the physical phenomenon of atmospheric dispersion of the sunlight most well known as the Rayleigh scattering effect. This technique does not depend on special algorithms to classify or to form clusters of objects appearing in complex aerial images. In fact, the mathematical model, derived from the physical model is very simple but it produces consistent results and can be applied to color images obtained by airborne and orbital sensors (e.g., Landsat TM and IKONOS).

The weighted gradient: a color image gradient applied to morphological segmentation

This paper proposes a method for color gradient computation applied to morphological segmentation of color images. The weighted gradient (with weights estimated automatically), proposed in this paper, applied in conjunction with the watershed from markers technique, provides excelent segmentation results, according to a subjective visual criterion. The weighted gradient is computed by linear combination of the gradients from each band of an image under the IHS color space model. The weights to each gradient are estimated by a systematic method that computes the similarity between the image to compute the gradient and an ”ideal image”, whose histogram has an uniform distribution. Several experiments were done in order to compare the segmentation results provided by the weighted gradient to the results provided by other color space metrics, also according to a subjective criterion, and such comparison is present in this paper.

Enhancement of Terrestrial Surface Features on High Spatial Resolution Multispectral Aerial Images

This work introduces new transformation indexes, color and spectral-based ones, applied for features enhancement registered in aerial multispectral images acquired with high spatial resolution by the use of the newest technology in imagery airborne sensors as the digital photogrammetric cameras IGN and HRSC-AX. The new transformation indexes proposed in this work are: the normalized Shadow Index (nSI) and the spectral Shadow Index (sSI) for shadow enhancement; the Weighted Water Index (WWI) and the Weighted Water-Soil Index (WWSI) for water bodies enhancement and, the Road Water intensity Shadow Index (RWSI) for paved-road, water bodies and intensity shadowed areas enhancement.

USO DE SOMBRAS NA CORREÇÃO DO EFEITO VINHETE

Variacoes do valor de brilho de alvos da superficie podem ocorrer no registro realizado por sensores imageadores. A variacao desses valores de brilho que ocorrem em funcao da distância do pixel ao centro da imagem e conhecida como efeito vinhete e sua correcao tem como objetivo oferecer um ajuste de brilho homogeneo na imagem. A proposta deste trabalho e apresentar uma metodologia para determinar um modelo de minimizacao desse efeito vinhete com base numa funcao ajustada pelo metodo dos minimos quadrados, a partir dos valores de brilho de sombras. Uma das hipoteses adotadas e de que as variacoes de brilho registrados em pixels de sombra sejam adequadas para representar os efeitos do vinhete. Considerando que o efeito vinhete possa ser modelado como uma tendencia de variacao espacial dos valores de brilho da imagem foi realizada a analise de superficie de tendencia de um conjunto de pixels amostrados em regioes de sombras. Foram adotadas duas abordagens para representar os valores de sombras que foram processados. A primeira, baseada nos componentes R, G e B da imagem colorida; e a segunda estrategia, do componente I da imagem HSI. A fim de avaliar a metodologia realizou-se um estudo de caso com uma imagem aerea colorida. Os experimentos mostraram que o melhor resultado foi obtido pela aplicacao do modelo nas componentes RGB, permitindo concluir que o efeito vinhete pode ser modelado com base em superficies de tendencia ajustadas aos valores RGB das sombras.

AntSeg: The Application of Ant Colony Optimization to Interactive Image Segmentation

The computational process of image segmentation performs the partitioning of it in disjoint regions by the specification and application of specific criteria. This process can be automatic or interactive. The interactive process allows the user interference in this process by means of the insertion of new criteria at any instant of the process execution. This new criterion can lead to the reaching of more appropriate results for the solution to the problem. However, interactive process application requires the user to have a high ability, context knowledge and accuracy executing the tasks. The interactive process can include facilities by allowing that some user actions may be executed with relative accuracy. Using computational resources, such task is made easier, since a computer can perform with more accuracy and speed, from the guides interactively defined by the user. The literature presents many methods that perform this task, among them the live wire, intelligent scissors and the interactive watershed with markers. The first two use a minimum cost system, in other words, optimization, to their working. This paper proposes an interactive method applied to image segmentation that combines the artificial ant colonies behavior and the image gradient technique. The gradient, supplied by heuristic rules, is the surface the ants search the path to the desired solution. The segmentation starts with the imposition of an initial point at the border of the object of interest and, by the user adding new points along its border, new segments belonged to object border are conquered. The proposed method shows a high repeatability of the segmentation task.