There is anecdotal evidence that luminance dominates chromatic information in the perception of edges. As a result, a blurred chromatic edge appears less so when overlaid on a sharp luminance edge. This effect has not been quantified and may be caused by lower chromatic acuity or by differences in contrast or spatial frequency content between the channels, rather than a mechanism which is particularly sensitive to luminance edges. We investigated this in a series of experiments in which the chromatic and luminance components of photographic images were separated and Gaussian blurred to varying degrees in accordance with a staircase procedure. A two-interval-forced-choice procedure was used to measure participants’ blur detection thresholds for each channel (colour and luminance) with and without its sharp counterpart. Acuity for chromatic blur was poorer in general and reduced further in the presence of sharp luminance information. However, luminance blur thresholds were unaffected by the presence of sharp chromatic information. These findings remained robust even when the contrast of the channels was equated. Natural images contain more high spatial-frequency luminance information than chromatic. To investigate whether this was responsible for the effect we reversed the colour and luminance channels. The images were converted into MB-DKL space (Derrington, Krauskopf, & Lennie, 1984; Macleod & Boynton, 1979) and the LM and S channels were replaced with the luminance information and the luminance channel was replaced with half of the sum of the LM and S channels. Chromatic blur thresholds were then no longer poorer in general, but chromatic blur was still masked by luminance information and there was again no effect of chromatic information on luminance blur thresholds. It appears that the masking effect is specific to the luminance channel, suggesting that luminance information about edges is automatically used by the visual system in preference to colour.
Meeting abstract presented at VSS 2012