Non-Linear makes beautiful
Otherwise, nothing could be perceived. In this talk, I will review some of the essential features of the visual system, from retina to primary visual cortex, mainly from the point of view of signals and image processing. I will start with some elementary visual illusions that will be explained later through adaptive non-linear functions applied to the classical linear models of visual processing. We will see various applications of this principle, at the early level of the retina (like spatio-temporal adaptation to luminance, contrast and colour) leading to the extraction of intrinsic features of objects, and also at the level of the V1 cortex, leading to the classical after-effects and to scene categorization. Key-words: Retina, luminance contrast and colour constancy, adaptive non-linearity, focus/context, scene categorization, Curvilinear Component Analysis.