Browse Prior Art Database

Simulating Visual Performance Based on Basic Visual Processes

IP.com Disclosure Number: IPCOM000005181D
Publication Date: 2001-Aug-16

Publishing Venue

The IP.com Prior Art Database

Abstract

The understanding of human visual image processing can be useful to the evaluation of visually effective image enhancements, to the prediction of visible artifacts in hardcopy prints, and to the estimation of visible artifacts in compressed images. This report describes our effort in the development of a stimulus-driven vision model. Differing from a common approach of using a contrast-sensitivity-function (CSF) as a front-end filter, the approach described in this report is based on basic visual processes. They include a front-end low-pass filter, cone-type-specific local light adaptation in the retina, frequency specific masking in the cortex, and particularly the contribution of a DC component in the stimuli. The model has successfully simulated basic visual performance, such as CSFs at different luminance, temporal frequencies, and field size, and can simulate the visible contrast contour of a frequency-modulated grating.