Abstract
Past studies in structure from motion(SfM) revealed that different depth cues such as stereoscopic disparities disambiguate interpretation of SfM. The present study investigated the effect of adding two types of motion cues, shearing motion (e.g. Rogers and Graham, 1979) and compressive motion (e.g. Ullman, 1979) for interpretation of SfM. The stimuli were random dot kinematograms. Relative velocity between dots includes both shearing and compressive components. Both components were projection of a rigid rocking motion (rotation). However, they could be interpreted separately ether rigid or nonrigid motion. Observers were asked to distinguish whether the stimulus appeared rigid or not. In the first experiment, average velocity for shearing motion was fixed and that for compressive motion was manipulated. In the second experiment, averaged velocity for compressive motion was fixed, and that for shearing motion was manipulated. Observers perceived mainly nonrigid motion when the stimuli included only one component, regardless of motion types. However, they perceived rigid rotation as the other motion component increased. This tendency was found in both experiments. These results indicate that the mechanism for SfM utilizes both shearing and compressive motions, and perceptual rigidity increases as variation of cues increases, even when they are only motion components.