Colour by correlation in a three dimensional colour space

We improve the promising Colour by Correlation method for computational colour constancy by modifying it to work in a three dimensional colour space. The previous version of the algorithm uses only the chromaticity of the input, and thus cannot make use of the information inherent in the pixel brightness. Our work is thus motivated by two observations. First, based on pervious work with the related gamutmapping class of algorithms, we know that the pixel brightness can make a substantial contribution to colour constancy, even if only the chromaticity of the illuminant is sought. Second, the success of the initial version of Colour by Correlation indicates that its use of statistical knowledge about the world is also useful for colour constancy. We develop the algorithm for the Mondrian world (matte surfaces), the Mondrian world with fluorescent surfaces, and the Mondrian world with specularities. We also point out how to extend the algorithm to take into account metallic specularities. The statistical occurrence of each of these types of surfaces in the real world can be modeled by the algorithm, which is significant because, ideally, colour constancy algorithms should be ready to deal with a wide variety of input, even though specific instances of a specific type of input are uncommon. We test the new algorithm on synthetic data, and on a data set of 321 carefully calibrated images. We find that on the synthetic data, the new algorithm significantly out-performs all other colour constancy algorithms. In the case of image data, the results are also promising. The new algorithm does significantly better than its chromaticity counter-part, and its performance approaches that of the best algorithms. Since the research into the method is still young, we are hopeful that t..