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.

This text was extracted from a WORD97 document.
This is the abbreviated version, containing approximately 5% of the total text.

Simulating Visual Performance Based on Basic Visual Processes

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 sizes, and can simulate the visible contrast contour of a frequency-modulated grating.

INTRODUCTION

As research in human vision provides a better understanding of human visual processing and computing power increases, the ability to apply knowledge of human visual processing to engineering applications through the application of vision models improves. These vision models can be applied in many applications, including estimating visible artifacts in compressed images, predicting the visibility of artifacts in hardcopy prints, evaluating the visibility of image enhancements, and designing visual displays that match visual performance.

On the other hand, it is necessary to recognize that our current understanding is limited. We can not satisfactorily explain many visual phenomena with our current understanding, especially when these phenomena are mediated by high level visual perception. However, a reasonable level of knowledge does exist regarding lower level visual processing (i.e., sensory visual information processing). This work is an attempt to create a model that predicts lower level processing and aims to develop a stimulus-driven vision model of image processing.

Introduction to Visual Processing Models. In a broad sense, there are two major issues that must be overcome to create a successful visual model. One issue is the ability to predict the capacity of information capture in the visual system, which determines the degree of fine spatial structure and color information that can be utilized by the visual system. These properties can be modeled simply by using a low-pass filter and a color-encoding scheme with three primaries. The second issue, which has consumed considerable effort in the visual modeling community, is the development of an approach to model the nonlinear processing in the visual system, such as light adaptation and frequency masking. It is important to note that the effects of the nonlinear processes are confined to local areas of a conce...