imcolordiff

Color difference based on CIE94 or CIE2000 standard

Since R2020b

Syntax

dE = imcolordiff(I1,I2)

dE = imcolordiff(I1,I2,Name=Value)

Description

dE = imcolordiff(I1,I2) calculates the color difference between color images or two sets of colors using the CIE94 standard. By default, the function expects the colors to be in the sRGB color space.

dE = imcolordiff(I1,I2,Name=Value) specifies additional aspects of the computation, such as the input color space and the CIE standard, using one or more name-value arguments.

Examples

Calculate Color Difference of Images Using CIE94 Standard

Read a color image into the workspace.

I1 = imread('peppers.png');
imshow(I1)

Alter the local color contrast in the image.

I2 = localcontrast(I1);
imshow(I2)

Calculate the color difference of the images using the default color standard, CIE94.

dE = imcolordiff(I1,I2);

Display the color difference as an image. Scale the display range to use the full range of pixel values in dE.

imshow(dE,[])

Calculate Color Difference of L*a*b* Images using CIE94 Standard

Read and display an image of tissue stained with hematoxylin and eosin (H&E).

he = imread("hestain.png");
imshow(he)

Convert the image to the L*a*b* color space.

lab = rgb2lab(he);

Make a copy of the image, then increase the signal of the a* channel. Red tones in the image become more saturated while the image overall brightness and the blue tones are unchanged.

lab2 = lab;
scaleFactor = 1.1;
lab2(:,:,2) = scaleFactor*lab(:,:,2);

Calculate the color difference of the original and enhanced image in the L*a*b* color space.

dE = imcolordiff(lab,lab2,isInputLab=true);

Display the color difference as an image. Scale the display range to match the range of pixel values in dE. Bright regions indicate the greatest color difference and correspond with the pink regions of tissue.

imshow(dE,[])

Calculate Color Difference of Two Colors using CIEDE2000 Standard

Specify two RGB color values.

pureRed = uint8([255,0,0]);
darkRed = uint8([255,10,50]);

Calculate the color difference of the colors using the CIEDE2000 standard.

dE = imcolordiff(pureRed,darkRed,"Standard","CIEDE2000")

dE = single

7.4449

Calculate Color Difference using Textile Weighting Factors

Read and display an RGB image of fabric.

fabric = imread('fabric.png');
imshow(fabric)

Simulate a second image of fabric by altering the local color contrast in the image.

fabric2 = localcontrast(fabric);
imshow(fabric2)

Calculate the color difference of the two images using the CIEDE2000 standard. Specify a luminance coefficient and K1 and K2 weighting factors appropriate for textiles.

dE = imcolordiff(fabric,fabric2,'Standard','CIEDE2000', ...
    'kL',2,'K1',0.048,'K2',0.014);

Display the color difference. Scale the display range to the full range of pixel values in dE.

imshow(dE,[])

Input Arguments

I1 — First set of color data

1-by-3 numeric vector | c-by-3 numeric matrix | numeric array

First set of color data, specified in one of these formats.

• 1-by-3 numeric vector representing a reference color.

• c-by-3 numeric matrix representing a set of c colors.

• m-by-n-by-3 numeric array representing a color image.

• Multidimensional numeric array, such as an m-by-n-by-3-by-p array, representing a batch of color images. The third dimension must correspond to the color channel and have 3 channels.

If I2 is not a reference color, then I1 must be either a reference color or a numeric array of the same size as I2.

I1 and I2 must be in the same color space. By default, the imcolordiff function expects that I1 and I2 are in the sRGB color space. If I1 and I2 are instead in the L*a*b* color space, then specify the isInputLab argument as true. L*a*b* color values can be of data type single or double only.

Data Types: single | double | uint8 | uint16

I2 — Second set of color data

1-by-3 numeric vector | c-by-3 numeric matrix | numeric array

Second set of color data, specified in one of these formats.

• 1-by-3 numeric vector representing a reference color.

• c-by-3 numeric matrix representing a set of c colors.

• m-by-n-by-3 numeric array representing a color image.

• Multidimensional numeric array, such as an m-by-n-by-3-by-p array, representing a batch of color images. The third dimension must correspond to the color channel and have 3 channels.

If I1 is not a reference color, then I2 must be either a reference color or a numeric array of the same size as I1.

I1 and I2 must be in the same color space. By default, the imcolordiff function expects that I1 and I2 are in the sRGB color space. If I1 and I2 are instead in the L*a*b* color space, then specify the isInputLab argument as true. L*a*b* color values can be of data type single or double only.

Data Types: single | double | uint8 | uint16

Name-Value Arguments

Specify optional pairs of arguments as Name1=Value1,...,NameN=ValueN, where Name is the argument name and Value is the corresponding value. Name-value arguments must appear after other arguments, but the order of the pairs does not matter.

Example: dE = imcolordiff(I1,I2,Standard="CIEDE2000") calculates the color difference between two RGB images using the CIEDE2000 standard.

Before R2021a, use commas to separate each name and value, and enclose Name in quotes.

Example: dE = imcolordiff(I1,I2,"Standard","CIEDE2000") calculates the color difference between two RGB images using the CIEDE2000 standard.

Standard — CIE standard

"CIE94" (default) | "CIEDE2000"

CIE standard used to compute the color difference value, specified as one of these values:

Value	Description
"CIE94"	The CIE94 standard. This standard improves the perceptual non-uniformities of the CIE76 standard implemented in the deltaE function.
"CIEDE2000"	The CIEDE2000 standard. This standard further improves the perceptual uniformity through five additional corrections: a hue rotation term, compensation for neutral colors, and compensation for lightness, chroma, and hue.

Data Types: char | string

isInputLab — Color values are in L*a*b* color space

false or 0 (default) | true or 1

Color values are in the L*a*b* color space, specified as a numeric or logical 0 (false) or 1 (true).

kL — Luminance coefficient

1 (default) | numeric scalar

Luminance coefficient, specified as a numeric scalar. kL is typically 1 for applications in graphic arts and 2 for applications in textiles.

kC — Chroma compensation coefficient

1 (default) | numeric scalar

Chroma compensation coefficient, specified as a numeric scalar. kC is typically 1.

kH — Hue compensation coefficient

1 (default) | numeric scalar

Hue compensation coefficient, specified as a numeric scalar. kH is typically 1.

K1 — K1 weighting factor

0.045 (default) | numeric scalar

K1 weighting factor, specified as a numeric scalar. The K1 weighting factor applies to the CIE94 standard only. The value is typically 0.045 for applications in graphic arts and 0.048 for applications in textiles.

K2 — K2 weighting factor

0.015 (default) | numeric scalar

K2 weighting factor, specified as a numeric scalar. The K2 weighting factor applies to the CIE94 standard only. The value is typically 0.015 for applications in graphic arts and 0.014 for applications in textiles.

Output Arguments

dE — Color difference

numeric matrix | c-by-1 column vector | numeric scalar | numeric array

Color difference (delta E), returned as one of the following.

• An m-by-n numeric matrix. The array represents the pixel-wise color difference between two color images, or between a color image and a reference color.

• A c-by-1 numeric column vector. The vector represents the color difference between two sets of c colors, or between a set of c colors and a reference color.

• A numeric scalar that represents the color difference between two reference colors.

• A multidimensional numeric array. The array represents the pixel-wise color difference between two batches of color images, or between a batch of color images and a reference color. The third dimension has length 1 and indicates the color difference.

If I1 or I2 is of data type double, then dE is of data type double. Otherwise, dE is of data type single.

Data Types: single | double

Tips

• To calculate color differences following the CIE76 standard, use the deltaE function. This function is faster than the imcolordiff function, but less precise.

• Typically, you should keep the default values of the kL, kC, kH, K1, and K2 arguments. The values can change for specific industry groups, such as graphic arts and textiles, based on experimentally validated lighting conditions for that industry.

References

[1] Sharma, Gaurav, Wencheng Wu, and Edul N. Dalal, "The CIEDE2000 Color-Difference Formula: Implementation Notes, Supplementary Test Data, and Mathematical Observations." Color Research and Application 30, no. 1 (February 2005): 21–30. https://doi.org/10.1002/col.20070.

[2] ISO/CIE 11664-6:2014. "Colorimetry — Part 6: CIEDE2000 Colour-difference formula." International Organization for Standardization. URL: https://www.iso.org/standard/63731.html.

[3] CIE 116-1995. "Industrial Colour-Difference Evaluation (E)." International Commission on Illumination (CIE). URL: https://cie.co.at/publications/industrial-colour-difference-evaluation.

Version History

Introduced in R2020b

R2023a: Support for reference colors and multidimensional images

You can now calculate the color difference between each pixel in an image (or each color in a set of colors) and a reference color. To do this, specify either I1 or I2 as a 1-by-3 numeric vector that represents the reference color.

You can now specify I1 and I2 as multidimensional numeric arrays representing batches of color images.

R2023a: New name-value arguments kC and kH

The kC and kH name-value arguments enable you to specify chroma compensation and hue compensation, respectively. These compensations are relevant for both the CIE94 and CIEDE2000 standards.