Abstract
Perceived lightness is often measured by asking observers to specify a test field that best matches their percept of a target. Typically one is interested in measuring lightness constancy, i.e. to what extent observers choose the same matches across viewing contexts that lead to variations in the luminance signal of the target. Such a procedure is called asymmetric matching which points to the potential problem that perceived lightness must be compared between different viewing contexts. The procedure of maximum-likelihood difference scaling (MLDS), on the other hand, allows the estimation of perceptual scales from judgments of stimulus differences along one stimulus dimension. That is, perceptual lightness scales can be obtained for lightness judgments performed within different viewing contexts and should theoretically map to the same internal scale, given lightness constancy. Here, we wanted to test this by measuring perceived lightness as a function of different viewing contexts in a matching and an MLDS procedure. Stimuli were rendered images of variegated checkerboards. Perceived target lightness was assessed for ten different reflectance values and five different viewing contexts (plain view and four transparencies). In the MLDS procedure we used a triad comparison, i.e. observers indicated which of two pairs of checks, presented in the same viewing context, appeared more different in lightness. In the matching procedure, observers adjusted an external test field to match the lightness of the target check. We found a high correspondence between the two procedures. In the MLDS procedure similar perceptual scales were estimated for each viewing context indicating a mapping to the same internal dimension. Simulations of MLDS scales on either luminance- or reflectance-based judgments showed that the observed scales were more consistent with reflectance-based judgments. As observers reported unanimously that the triad comparison was the more intuitive task, MLDS might provide an interesting alternative to standard matching procedures.
Meeting abstract presented at VSS 2016