Image information content assessment and elicitation

Abstract

Despite impressive progress in computer technologies, evolutionary shaped biological visual systems continue to be superior over human-made (computer-based) systems. One of the reasons for that is the striking dissimilarity in low-level image processing paradigms that the two systems sustain. While biological systems are busy with information processing, (from the very beginning, from the input front-end), their computer-based analogs are occupied with data processing, - endless and tireless number crunching. It is widely believed, that mimicking properties of biological visual systems in design of their computer-based counterparts will greatly improve the performances of the latter. Although convincing evidences of such ventures are still, let us say, not widely known, this mock-up approach is exactly what we had attempted to carry out, and what we would like to explain in this paper.