When we observe moving objects behind some occluders and only motions of local parts are visible, the entire motion cannnot be perceived. This is called "aperture problem". This problem can be solved by integrating two or more local information (motion integration). Previous studies about motion integration usually used 2D stimuli. Only a few studies used 3D stimuli. Among then, there was the one which investigated the relation between collinear arrengement of moving lines in depth dimension and motion integration, and the relation between figure-ground separation and motion integration, using dynamic random dot stereogram. This study performded psychophysical experiments to clarify the relation between figure-ground separation and motion integration in detail using 2D stimuli expressing concave/convex edges based on the charecteristics of spatiotemporal frequency of 2D pictures for the figure-ground separation, and 3D stimuli by dynamic randomdot stereogram. We obtained the results that when motions of concave edges were perceived, subjects could easily integrate them, on the other hand, when convex edges were presented, subjects could not integrate them well. This result suggests that there is a relation between figure-ground separation and motion integration.