Color constancy: generalized diagonal transforms suffice

Abstract

This study's main result is to show that under the conditions imposed by the Maloney-Wandell color constancy algorithm, whereby illuminants are three dimensional and reflectances two dimensional (the 3-2 world), color constancy can be expressed in terms of a simple independent adjustment of the sensor responses (in other words, as a von Kries adaptation type of coefficient rule algorithm) as long as the sensor space is first transformed to a new basis. A consequence of this result is that any color constancy algorithm that makes 3-2 assumptions, such as the Maloney-Wandell subspace algorithm, Forsyth's MWEXT, and the Funt-Drew lightness algorithm, must effectively calculate a simple von Kries-type scaling of sensor responses, i.e., a diagonal matrix. Our results are strong in the sense that no constraint is placed on the initial spectral sensitivities of the sensors. In addition to purely theoretical arguments, we present results from simulations of von Kriestype color constancy in which the spectra of real illuminants and reflectances along with the human-conesensitivity functions are used. The simulations demonstrate that when the cone sensor space is transformed to its new basis in the appropriate manner a diagonal matrix supports nearly optimal color constancy.

Seeing the light: Illumination as a contextual cue to color choice behavior in bumblebees

“…A widespread notion in vision research is that animals do this by recalibrating their visual system in such a way that the information about object illumination is itself discounted (12)(13)(14)(15)(16). In this view, deviations from perfect color constancy are the result of mechanistic limitations, and the visual system should interpret them as errors: if a change in illumination results in a minor change of object appearance, then animals would choose targets that most closely resemble those that had previously been entrained (typically referred to as imperfect color constancy).…”

“…source (5), or by differentially adapting the receptors to the average spectral power in the stimulus (12)(13)(14)(15)(16). Although each strategy has its advantages and disadvantages, a necessary corollary of this supposition is that higher-order processing and perception should be blind to changes in illumination, because the relevant information is ultimately lost.…”

Robust real-time detection of multi-color markers on a cell phone

“…It is well known [9] that if M s = 3, the relationship between the color of a surface s when seen under two different illuminants can be expressed as follows:…”

“…Likewise, under the diagonal model, the color values of two different surfaces s 1 and s 2 seen under the same illuminant i are related by a diagonal matrix that is independent of the illuminant: In fact, it holds true only if each matrix Q k has only one nonnull column. Finlayson et al [9] showed that when M i = 3 and M s = 2 (or M i = 2 and M s = 3), there exists an invertible linear color transformation such that the diagonal model holds exactly on the transformed colors.…”

A Composite Illumination Invariant Color Feature and Its Application to Partial Image Matching

“…The causes of such additional components and variations include the differences in object surface conditions and the spectral components of illuminations [2]- [4]. Modeling of perceptual color constancy also aims to achieve a similar goal [5], [6]. The photometric invariants have been used in 3D object recognition [7], correspondence, shape and motion estimation [4], and have achieved superior and robust results in comparison with the use of RGB color or grayscale.…”