Obtaining a color constant descriptor from the image pixels is not only important for digital photography but is also very important for computer vision. Many algorithms work only under one set of lighting conditions but not under another. For instance, an algorithm may work very well under natural lighting but the same algorithm may not work as well when used under artificial illumination. Color constant descriptors are also very important for color-based object recognition. At present, it is not known how color constant descriptors are computed by the human visual system. However, a number of algorithms have been proposed to address the problem of color constancy. The book describes all of the major color constancy algorithms that are known from the literature together with recent research done by the author.
Human color perception is only approximately constant as you probably have noticed when buying clothes in a store. If you select a set of seemingly black trousers you may very well find out at home that the selected set of trousers is actually kind of bluish. Color perception is also influenced by the colors which are present in the surround of an object. In this case, color perception could have been influenced by the lack of a sufficiently complex surround. You can find out the color of a set of trousers by putting the trousers next to another set of trousers. If you place a seemingly black set of trousers next to another one you may find out that one is actually kind of dark bluish whereas the other one is indeed black. If you take the black trousers and place them next to black velvet you will think that the trousers are kind of dark grey and that the velvet is actually black. Why color perception sometimes behaves as just described will become clearer after reading this book.