How can I decrease image contrast using simple arithmetic?

Hi everyone,

I am quite new to Matlab and I am in need of help to find ways to decrease the contrast of an image:

a) mathematically without using functions such as imadjust, etc. b) while keeping the average luminance the same

Here is what I have been using:

 im = double(im);
    R = im(:,:,1);
    G = im(:,:,2);
    B = im(:,:,3);
    im = 0.2989*R + 0.5870*G + 0.1140*B;
    im1 = (im/255);
contrast_factor = 0.6;
av_lum = mean(im1(:));
scaled_image = ((im1 - av_lum)*contrast_factor) + av_lum

This seems to be working but I am not sure if it is the most accurate way. Do I need to use the round function? I will be using this code on at least 60 images with different histograms.

Any help would be greatly appreciated!

Answer 1

[David Young]

Your code looks fine.

You should almost certainly not use the round function. If your image started off as values in the range 0-255, im1 will have values in the range 0-1, so the round function will destroy most of the information.

You are making the assumption that luminance is linearly related to pixel value. That's probably alright, though there are applications where it may not be what you want. We'd need more information to comment on that.

[Lauren]

Does gamma correction allow for a linear relationship between luminance and pixel value? The monitor I will be using will be gamma corrected.

[Image Analyst]

Gamma FAQ: http://www.poynton.com/GammaFAQ.html

Answer 2

[Image Analyst]

If you want it to be faster you can compute the look up table from your equation, and then use intlut().

Answer 3

[tom]

Hey everyone,

sorry for the "add-on question", but I try to do the same as Lauren ( (1.) reduce image contrast & (2.) keep pixel intensity constant using simple arithmetric). But in the end I would like to "keep" or recover the RGB/Color image. In Laurens aprroach the RGB image is transformed into grayscale. I dont know if this even possible?

Any help or recommendation would be greatly appreciated!

Tom

[Image Analyst]

In Lauren's code, the im variable is not still available because she overwrite it. If you want to save it, don't overwrite it - just create new R, G, and B variables.

%im = double(im); % Don't do anything at all to im to leave it as the original RGB image.
R = double(im(:,:,1));
G = double(im(:,:,2));
B = double(im(:,:,3));
im1 = (0.2989*R + 0.5870*G + 0.1140*B) / 255;
contrast_factor = 0.6;
averageLuminance = mean(im1(:));
scaled_image = ((im1 - averageLuminance) * contrast_factor) + averageLuminance;

[tom]

Ah sorry, I did not express myself very well! The variable "scaled_image" in Laurens code appears to be grayscale at the end. But I want to reduce the contrast of the RGB image in a way that its still an color image at the end. I tried it using this code but I´m not completely sure about it:

im = double(img);

R = im(:,:,1); G = im(:,:,2); B = im(:,:,3);

im = 0.2989*R + 0.5870*G + 0.1140*B;

R1=R/255;

G1=G/255;

B1=B/255;

contrast_factor = 0.65;

av_lum1 = mean(R1(:));

scaled_image1 = ((R1 - av_lum1)*contrast_factor) + av_lum1;

av_lum2 = mean(G1(:));

scaled_image2= ((G1 - av_lum2)*contrast_factor) + av_lum2;

av_lum3 = mean(B1(:));

scaled_image3 = ((B1 - av_lum3)*contrast_factor) + av_lum3;

Adjusted_Img=cat(3,scaled_image1,scaled_image2,scaled_image3);

Thanks again!

[Image Analyst]

Yes, you'd have to do each channel one at a time. Instead of

R = im(:,:,1); G = im(:,:,2); B = im(:,:,3);

you can use

% Extract the individual red, green, and blue color channels using imsplit() (introduced in R2018b).
[R, G, B] = imsplit(im);

[tom]

thank you very much!