Modeling Visual Information Processing in Brain: A Computer Vision Point of View and Approach

摘要

Image understanding and image semantics processing have recently become an issue of critical importance in computer vision R&D. Biological vision has always considered them as an enigmatic mixture of perceptual and cognitive processing faculties. In its impetuous and rash development, computer vision without any hesitations has adopted this stance. I will argue that such a segregation of image processing faculties is wrong, both for the biological and the computer vision. My conjecture is that images contain only one sort of information-the perceptual (physical) information, which can be discovered in an image and elicited for further processing. Cognitive (semantic) information is not a part of image-conveyed information. It belongs to a human observer that acquires and interprets the image. Relying on a new definition of "information", which can be derived from Kolmogorov's complexity theory and Chaitin's notion of algorithmic information, I propose a unifying framework for visual information processing, which explicitly accounts for perceptual and cognitive image processing peculiarities. I believe, it would provide better scaffolding for modeling visual information processing in human brain.