Spatial standard observer

• US 7,783,130 B2
• Filed: 01/24/2005
• Issued: 08/24/2010
• Est. Priority Date: 01/24/2005
• Status: Active Grant

• Current Assignee: USA As Represented By The Administrator Of The NASA
• Original Assignee: United States Of America As Represented By The Administrator Of The National Aeronautics And Space Administration
• Inventors: Watson, Andrew B.
• Primary Examiner(s): Le; Vu
• Assistant Examiner(s): Abdi; Amara
• Application Number:
  US11/045,041
• Publication Number:
  US 20060165311A1
• Time in Patent Office: 2,038 Days
• Field of Search: 382/106, 382/263, 382/286, 382/279, 382/130
• US Class Current: 382/286
• CPC Class Codes: G06T 2207/30168 Image quality inspection
• CPC Class Codes: G06T 7/0002 Inspection of images, e.g. flaw detection
• CPC Class Codes: H04N 17/04 for receivers
• CPC Class Codes: H04N 19/60 using transform coding
• CPC Class Codes: H04N 19/86 involving reduction of coding artifacts, e.g. of blockiness

Abstract

The present invention relates to devices and methods for the measurement and/or for the specification of the perceptual intensity of a visual image, or the perceptual distance between a pair of images. Grayscale test and reference images are processed to produce test and reference luminance images. A luminance filter function is convolved with the reference luminance image to produce a local mean luminance reference image. Test and reference contrast images are produced from the local mean luminance reference image and the test and reference luminance images respectively, followed by application of a contrast sensitivity filter. The resulting images are combined according to mathematical prescriptions to produce a Just Noticeable Difference, JND value, indicative of a Spatial Standard Observer, SSO. Some embodiments include masking functions, window functions, special treatment for images lying on or near borders and pre-processing of test images.

Claims

1. A method of reducing wrap-around in a digital image processing process that compares a digital test image and a digital reference image and that includes one or more convolution steps, the method comprising:

a) producing a reference luminance image from the reference image and a test luminance image from the test image;

b) producing a local mean luminance reference image as a convolution of the reference luminance image and a luminance filter function;

c) producing a test contrast image that is at least one of the following;

(c1) a mathematical combination of the test luminance image and the local mean luminance reference image and (c2) a mathematical combination of the test luminance image, the local mean luminance reference image and a border aperture function and (c3) a mathematical combination of the test luminance image, the local mean luminance reference image and an image of a border surrounding the reference image;

d) producing a reference contrast image that is at least one of the following;

(d1) a mathematical combination of the reference luminance image and the local mean luminance reference image and (d2) a mathematical combination of the reference luminance image, the local mean luminance reference image and the border aperture function;

e) applying a contrast sensitivity filter to the test contrast image to produce a filtered test image;

f) applying the contrast sensitivity filter to the reference contrast image to produce a filtered reference image;

g) providing a difference image by at least one of the following two processes;

(g1) subtracting the filtered reference image from the filtered test image to produce a difference image, and (g2) producing a mask image as a mathematical combination of the filtered reference image with a masking filter, and producing a difference image as a ratio of the difference image and the mask image;

h) producing a just noticeable difference image as a mathematical combination of the difference image with a window function; and

i) pooling the just noticeable difference image to produce a visibility metric,

wherein the convolution operation in at least one of process (b) and process (h) is performed using a confined convolution process that comprises;

(i-1) receiving an image I(x,y), expressed as an array of k_xo-by-k_y0 pixels in an x-direction and in a y-direction, respectively, where k_x0 and k_y0 are selected positive integers;

(i-2) padding the image I(x,y) with k_x zeroes in the x-direction and by k_y zeroes in the y-direction, to provide a first intermediate image I1(x,y), expressed as an array of (k_x0+k_x)-by-(k_y0+k_y) pixels in the x-direction and in the y-direction, respectively, where k_x and k_y are selected non-negative integers;

(i-3) convolving the first intermediate image I1(x,y) with a selected non-negative kernel function K(x,y), expressed as an array of k_x-by-k_y pixels in the x-direction and in the y-direction, respectively;

, to obtain a second intermediate image I2(x,y) expressed as an array of (k_x0+k_x)-by-(k^y0+k_y) pixels in the x-direction and in the y-direction, respectively; and

(i-4) cropping the second intermediate image I2(x,y) to an array of k_x0-by-k_y0 pixels in the x-direction and in the y-direction, respectively, to obtain a third intermediate image I3(x,y), expressed as an array of k_x0-by-k_y0 pixels in the x-direction and in the y-direction, respectively.

2. The method of claim 1, further comprising preprocessing at least one of said test image and said reference image by downsampling.

3. The method of claim 1, further comprising preprocessing at least one of said test image and said reference image by convolution with a selected pre-filtering function.

4. The method of claim 1, further comprising preprocessing, by cropping, at least one of said test image and said reference image.

6. The method of claim 1, further comprising receiving said third intermediate image, I3(x,y)=PCC{K(x,y), I(x,y)} and forming a fourth intermediate image, defined as

I4(x,y)=K(x,y)©

I(x,y)

=PCC{K(x,y), I(x,y)}/PCC{K(x,y)/Σ

_x′

Σ

_y′

K(x′

,y′

), I(x,y)}.

5. A method of forming a digital reference image from a digital test image, the method comprising:

a) providing a test image having first and second opposing sides of the image

b) performing a confined convolution of the test image with a selected filter that isolates the first and second opposing sides of the image from each other, to thereby form the digital reference image,

wherein the confined convolution process comprises an ordered sequence of operations, denoted PCC, on an image, the sequence comprising;

(i-1) receiving an image I(x,y), expressed as an array of k_x0-by-k_y0 pixels in an x-direction and in a y-direction, respectively, where k_x0 and k_y0 are selected positive integers;

(i-2) padding the image I(x,y) with k_x zeroes in the x-direction and by k_y zeroes in the y-direction, to provide a first intermediate image I1(x,y), expressed as an array of (k_x0+k_x)-by-(k_y0+k_y) pixels in the x-direction and in the y-direction, respectively, where k_x and k_y are selected non-negative integers;

(i-3) convolving the first intermediate image I1(x,y)with a selected non-negative kernel function K(x,y), expressed as an array of k_x-by-k_y pixels in the x-direction and in the y-direction, respectively;

, to obtain a second intermediate image I2(x,y) expressed as an array of (k_x0+k_x)-by-(k_y0+k_y) pixels in the x-direction and in the y-direction, respectively; and

(i-4) cropping the second intermediate image I2(x,y) to an array of k_x0-by-k_y0 pixels in the x-direction and in the y-direction, respectively, to obtain a third intermediate image I3(x,y)=PCC {K(x,y), I(x,y)}, expressed as an array of k_x0-by-k_y0 pixels in the x-direction and in the y-direction, respectively.

7. The method of claim 5, further comprising receiving said third intermediate image, I3(x,y)=PCC{K(x,y), I(x,y)} and forming a fourth intermediate image, defined as

I4(x,y)=K(x,y)©

I(x,y)

=PCC{K(x,y), I(x,y)}/PCC{K(x,y)/Σ

_x′

Σ

_y′

K(x′

,y′

), I(x,y).