Uma modelagem computacional de áreas corticais do sistema visual humano associadas à percepção de profundidade

• Autor(a) principal: Oliveira, Hallysson
• Data de Publicação: 2005
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Repositório Institucional da Universidade Federal do Espírito Santo (riUfes)
• Texto Completo: http://repositorio.ufes.br/handle/10/6350
• topic: Ciência da Computação; Profundidade - Percepção; Visão por computador; Modelagem de dados; 004

The images formed on the human retinas has only two dimensions, however the brain is capable of synthesizing a three-dimensional representation with color, shape and depth information about the environment surroundings. For in such a way, after choosing a point in the 3-D space, the eyes verge to this fixation point and, at the same time, the visual system is fed back with eyes placement information, supplied by ocular motor system, considering them as the distance of this point to the observer. The depth perception around the vergence point may be got by estimating the disparity, the positional difference between the two retinal projections of a given point in space results from the horizontal separation of the eyes. Most of the depth perception processing is done by visual cortex, mainly in V1 and MT areas. This work presents a modeling and implementation of a mathematical computational neural architecture that represents the V1 and MT cortical areas, based in visual cortex neurophysiology, used to create in a computer a 3-D representation of the surrounds, using images from two cameras connected to this computer. This model implementation was based upon Gabor filters and log-polar mapping. With this model it is possible to make a disparity map of the images and thus calculate the 3-D spatial position of each image point represented in the disparity map. This way it is possible to create a 3-D representation of the surrounds in a computer.