Abstract
Most of our understanding of color difference perception is grounded in studies of colored patches of uniform size and shape. However, our ability to distinguish between colors depends on the spatial properties of objects being compared. This work statistically models how variations in size and shape change people's abilities to perceive color differences between two objects. Our models describe color difference perception as a function of how frequently people can correctly identify a difference between patches at fixed color distances for three patch shapes---circles (as in points on a scatterplot), lines (as in lines in a line graph), and rectangles (as in bars on a bar chart)---ranging in size from 0.125 to 6 degrees. We derive these models from a set of binary forced-choice comparison experiments conducted using 378 participants across 29,862 trials. These experiments presented participants with a sequence of paired colored patches represented as marks on a visualization. Each pair of patches had a known width and height and a controlled color distance between them. We found an inverse relationship between color discriminability and diameter for circular patches and between discriminability and line height for lines. Our results indicate that color differences are more readily perceived on lines than on circles of equivalent height; however, these gains are not proportional to the corresponding area differences between these patches. Instead, discriminability for rectangles and lines varies according to a weighted combination of the length of the longest edge and the ratio between edge lengths. Our results provide evidence of a direct correlation between the spatial properties of objects and their perceived colors. The corresponding models can be used to predict how the visual system's ability to distinguish colors changes based on the size and dimensions of objects being compared.
Meeting abstract presented at VSS 2017