Abstract
[Purpose] The purpose of this study was to investigate whether the recovery of 3-D structure from motion (SFM) was influenced when image features were segregated on the basis of a color difference. [Methods] We used two-cylinder SFM stimuli like those used by Ramachandran et al. (1988) in which two cylinders of the same diameter simulated with random dots on their surfaces were superimposed on each other, but rotated at different speeds. The perceived curvature of each cylinder surface was measured with a matching method when the two groups of dots belonging to different cylinders had the same color (unsegregated condition) and when they had different colors (red vs. green, segregated condition). Relative rotating speeds of two cylinders were systematically varied. [Results] The results for the unsegregated condition indicated that the observers perceived two distinct cylinders with the faster cylinder appearing more convex than the slower one, which quantitatively confirmed Ramachandran et al.'s demonstration. The results also showed that the difference in perceived curvature increased monotonically with the difference in rotating speed of two cylinders. When two groups of dots on different cylinders were differentiated by a color difference in the segregated condition, the difference in perceived curvature was greatly reduced compared with the unsegregated condition. Thus the surfaces of two rotating cylinders appeared more closely located in 3-D space. Interestingly, the influence of color segregation disappeared when the cylinders were of different diameters and thus two distinct cylinders could be easily recovered. [Discussion] The present results imply that color segregation can modulate some aspect of SFM processing, possibly the one assigning depth values to the surfaces according to the differences in speed of image features.