On the behavior of spatial models of color

There is a growing family of algorithms that treat/modify/enhance color information in its visual context, also known as Spatial Color methods (e.g. Retinex or ACE). They produce results that, due to a changing spatial configuration, can have a non-unique relationship with the physical input. In authors' opinion judging their performance is a challenging task and is still an open problem. Two main variables affect the final result of these algorithms: their parameters and the visual characteristics of the image they process. The term visual characteristics refers not only to the image's digital pixel values, (e.g. calibration of pixel value, the measured dynamic range of the scene, the measured dynamic range of the digital image), but also to the spatial distribution of these digital pixel values in the image. This paper does not deal with tuning parameters, rather it discusses the visual configurations in which a Spatial Color methods show interesting, or critical behavior. A survey of the more significant Spatial Color configurations will be presented and discussed. These configurations include phenomena, such as color constancy and contrast. The discussion will present strengths and weaknesses of different algorithms, hopefully allowing a deeper understanding of their behavior and stimulating discussions about finding a common judging ground.