xiangpenghao / colorconstancy

Repo for color constancy algorithm comparison

Color Constancy Algorithms

Load a spectral image

img1 = SpectralImage.NewFromFile('path_to_img.exr') # load the exr image file
img2 = SpectralImage.NewFromSpectrum(np_ndarray) 
print(img[405]) # print out the spectral intensity at 405nm

Convert a spectral image to a sRGB image

rgb_img = spectral_img.to_srgb()
rgb_img.dump_file('output_rgb.exr')

Evaluate shades of grey

from benchmark import BaseBench
from algorithms.shades_of_gray import AverageRGB, MaxRGB, PNorm
from images import SpectralImage, sRGBImage

test_file = 'test1'

algs: [BaseBench] = [AverageRGB, MaxRGB, PNorm]
spectral_img = SpectralImage.NewFromFile(f'fixtures/{test_file}.exr')
rgb_img = spectral_img.to_srgb()

for algorithm in algs:
    refl = algorithm(rgb_img, None).get_test_reflectance()
    refl.dump_file(f'dist/{test_file}_{algorithm.NAME}_refl.exr')

colorconstancy's Issues

are the dataset available in the paper

Hi, thanks for your awesome work!
Are the dataset available in the paper "Evaluating Colour Constancy on the new MIST dataset of
Multi-Illuminant Scenes"?