Motion perception realized with dynamics of motion boundary

When we see a random-dot kinematogram (RDK for short) in which a small region moves on a background, we can perceive the small region separated from the background with a distinct contour. This separation is often called “pop out”. On the other hand, when we see an unusual RDK in which different motions are given in regions beside the motion boundary but there is no “motion” of motion boundary, we can hardly perceive the pop out phenomenon and a distinct contour. So, we suppose that it is very important for motion perception to consider the motion boundary. We introduce the effects of advection and diffusion into the dynamics of motion boundary, and indicate that consecutive patterns of motion with boundaries are estimated efficiently