Method and apparatus for determining visually perceptible differences between images

Claim 1

Apparatus for determining visual differences, to the human visual system of an observer, between a first image signal representative of a first image and having a first luminance level and a second image signal representative of a second image and having a second luminance level, comprising: an amplitude non-linearity unit, which is operative to adjust said first and second luminance levels of the first and second image signals, respectively; a contrast sensitivity unit, which is coupled to the output of said amplitude non-linearity unit, and is operative to determine variations in visual sensitivity of said first and second images as a function of frequency; a cortex unit, which is coupled to the output of said contrast sensitivity unit, and is operative to subdivide each of said first and second image signals into a plurality of cortex-filtered image signals which vary according to visual orientation of said first and second images relative to said human visual system of said observer and frequency content of image information; a contrast difference unit, which is coupled to the output of said cortex unit, and is operative to provide image signals indicative of the difference between said cortex-filtered first and second image signals at a plurality of visual orientation and frequency band combinations of said cortex filters; a first threshold elevation unit, which is coupled to receive a plurality of cortex-filtered first image signals from said cortext unit, and is operative to determine the visual sensitivity of first cortex-filtered images according to the frequency and orientation of information in the first image signal; a second threshold elevation unit, which is coupled to receive a plurality of cortex-filtered second image signals from said cortext unit, and is operative to determine the visual sensitivity of second cortex-filtered images according to the frequency and orientation of information in the second image signal; a threshold elevation comparison unit, which is coupled to the outputs of said first and second threshold elevation units, and is operative to determine an overall threshold according to first threshold elevations associated with said first plurality of cortex-filtered images and second threshold elevations associated with said second plurality of cortex-filtered images; and a probability generator unit, which is coupled to the output of aid contrast unit and the output of said threshold elevation comparison unit, and is operative to determine the probability that a user will visually perceive a difference between the first image and the second image.

Claim 10

A method of processing first and second digital images, each of said first and second digital images being comprised of a respective array of pixels, and having differences therebetween that are visually perceptible to a human observer, in order to selectively extract from each digital image a plurality of digital images comprising the steps of: (a) for each of said first and second digital images, modeling a non-linear response of cone receptors of the human eye to the luminance of each pixel of said each first and second digital image, to produce a respective non-linear response digital image; (b) processing each respective non-linear response digital image produced in step (a) in accordance with a contrast sensitivity function representative of variations in visual sensitivity as a function of spatial frequency, to produce a respective contrast sensitivity function digital image; and (c) processing each respective contrast sensitivity function digital image produced in step (b) in accordance with a cortex filter operator which is operative to adjust said each contrast sensitivity function digital image as a function of orientation and frequency content of a prescribed image feature and surrounding image content, so as to derive a plurality of digital images associated with respectively different frequency bands and orientations.

Claim 11

A method of processing first and second digital images, each of said first and second digital images being comprised of a respective array of pixels, and having differences therebetween that are visually perceptible to a human observer, in order to selectively extract from each digital image a plurality of digital images comprising the steps of: (a) for each of said first and second digital images, modeling a non-linear response of cone receptors of the human eye to the luminance of each pixel of said each first and second digital image, to produce a respective non-linear response digital image; (b) processing each respective non-linear response digital image produced in step (a) in accordance with a contrast sensitivity function representative of variations in visual sensitivity as a function of spatial frequency, to produce a respective contrast sensitivity function digital image; and (c) processing each respective contrast sensitivity function digital image produced in step (b) in accordance with a cortex filter operator which is operative to adjust said each contrast sensitivity function digital image as a function of orientation and frequency content of a prescribed image feature and surrounding image content, so as to derive a plurality of digital images associated with respectively different frequency bands and orientations.