Color Math and Programming

Question

I am currently trying to build a program that will accept a image and replace all of its individual parts 20x20 with individual images (up to a 1000) that are a best visual match.

I researched and learned about avg RGB/Color, my broad idea of how I would build the program is that I first calculate the average Color/RGB transform the RGB to CIE-L*ab and select the image (up to a 1000) that has the smallest difference to a 20x20 pixel of the input image.

I am not asking for code, just some resources that can better help me understand how I get the transformation from RGB to CIE-L*ab 1976 Comparison.

I am not interested in Euclidian distance to calculate the distances because it does not take human color perception into account. If it helps, i'm building the program in NodeJS

Thank You

I think https://github.com/toish/chromatism may be helpful to you. — user3351605, Mar 01 '18 at 13:35
@user3351605 Thanks, looks very helpful. I will make sure to take a look! — adot, Mar 01 '18 at 15:24
if that library ends up working well for your needs, consider creating a self-answer that explains how you were able to use it to solve your problem. — user3351605, Mar 01 '18 at 18:07

score 0 · Accepted Answer · answered May 11 '18 at 07:02

How I managed to do it was using two node modules, image-average-color and color-space. I first calculated the average color of each image in the collection, converted it to CIE-L*ab and stored it in an array.

I then extracted each 20x20 block of the input image, calculated the CIE-L*ab aswell and found the smallest distance in the array to the 20x20 block

Color Math and Programming

1 Answers1