Visual attention is the capability of biological visual systems to rapidly detect interesting parts of the retinal image, in order to reduce the amount of data for complex processing tasks, such as object recognition. In this thesis, a computational model of visual attention is described, which combines knowledge about the human visual system with the use of computer vision techniques. The model is structured into three major stages. First multiple feature maps are extracted from the input image (e.g. orientation, curvature, color contrast, etc). Second, a corresponding number of "conspicuity maps" are computed, which enhance region of interest in each feature map. Third a non-linear relaxation process is used to integrate the conspicuity maps into a single representation. Extensions of the bottom-up system are proposed. These include the use of a-priori knowledge about objects of interest for the generation of top-down attention information.