The Frequency of Metamerism in Natural Scenes

Abstract

Estimates of the frequency of metameric surfaces, which appear the same to the eye under one illuminant but different under another, were obtained from 50 hyperspectral images of natural scenes. The degree of metamerism was specified with respect to a color-difference measure after allowing for full chromatic adaptation. The relative frequency of metameric pairs of surfaces, expressed as a proportion of all pairs of surfaces in a scene, was very low. Depending on the criterion degree of metamerism, it ranged from about 10 −6 to 10 −4 for the largest illuminant change tested, which was from a daylight of correlated color temperature 25,000 K to one of 4000 K. But, given pairs of surfaces that were indistinguishable under one of these illuminants, the conditional relative frequency of metamerism was much higher, from about 10 −2 to 10 −1 , sufficiently large to affect visual inferences about material identity.

The Demands of Geometry on Color Vision

“…By the same token, the arguments here depend on the logic of plane geometry and not on any partiulcar metric of color space. Although the question remains, the metameric stimuli used in colorimetry experiments are rare in natural images [32]. Thus the evolution of color vision may simply have ignored the legitimate conceptual problem posed by metamers.…”

Performance evaluation of dimensionality reduction techniques for multispectral images

“…We investigate the following classes of spectral data: natural illumination spectra, our own indoor images of a static scene under different illumination conditions, and the reflectance images of indoor and outdoor scenes described in [4]. The investigation of different types of illuminations is not only an interesting research issue, but we also combine those different illuminations to create simulated images from the reflectance images.…”

“…The database described in [4] consists of 50 close-up and distant images of a great variety of urban and rural scenes from the Minho region of Portugal. These images were recorded in sunlight without clouds or under a sky with uniformly distributed clouds.…”

“…The effective spectral reflectance at each pixel is computed normalizing the signal against that obtained from the reference surface. For more details on the acquisition procedure, the reader is referred to [4]. In Figure 4 (a) we can see the image from the database that we selected for our experiments.…”

“…We can then use this to compensate the effects of the illumination change or generate the appearance of the scene under a different illumination (a detailed description of such an application is described in [17]). We apply these techniques to investigate different types of spectral data: analytically defined spectra (Plank spectra), D-type illuminations obtained by empirically derived approximations [9,28], a series of measurements of daylight spectra from Norrköping, Sweden and a series of multispectral images of indoor and outdoor scenes, some obtained with our own acquisition system and others from other research groups [4]. In the following section, we describe in detail the spectra used in the experiments.…”