adjust brightness of an image with respect to another image.

Is there some automatic function for that, something like imregister only to equalize the brightness between two images ? Thanks.

Answer 1

[Walter Roberson]

The below might not look like it makes the brightnesses of the images the same, but mathematically it does the job, provided that you do not get saturation.

b1 = mean2(rgb2gray(Image1));
b2 = mean2(rgb2gray(Image2));
NewImage2 = Image2 + (b1-b2);

[Elad]

Thanks, unfortunatly that does not work well in my case. perhaps I need some local equalization between the images, the algorithm by Mark Grundland and Neil A. Dodgson that Image Analyst suggested might work.

Answer 2

[Image Analyst]

For the ultimate in histogram matching of grayscale images, see my histogram matching/shaping routine. http://www.mathworks.com/matlabcentral/fileexchange/?term=authorid%3A31862 You can't do better since it will provide as exact a match as is theoretically possible. (It's only slightly inexact when the image sizes are different, for obvious reasons. Ask if you don't know why.)

For a color image, I think the algorithm by Mark Grundland and Neil A. Dodgson is possibly the best algorithm out there: http://www.eyemaginary.com/Portfolio/ColorHistogramWarp.html

[Elad]

Thanks, Im working with color images, is there an available implementation for Mark Grundland and Neil A. Dodgson algorithm? (Matlab or other)

[Image Analyst]

Not that I know of.

[Thang Le]

Sorry for digging up an old post. I've looked at the custom-shaped histogram package and I'm not sure how to equalize the brightness and contrast across multiple images. I have 160 images with different levels of brightness and contrast and I would like to create a set of new 160 images from these with the same level of brightness and contrast. Image Analyst, could you help?

[Image Analyst]

Try using imhistmatch() in a loop.

Answer 3

[px1313]

I would like to ask you if imhistmatch() have worked for your 160 images with different levels of brightness and contrast. I have same problem to solve.