Omni-directional image matching for homing navigation based on optical flow algorithm

It is known that many insects use optical flow as means of homing navigation. Even though insects have a relatively small number of neurons to estimate spatial information, compared with mammals, insects can maintain high performance of homing navigation to return home. This implies that a navigation model of insect vision using optical flow has high efficiency. There have been suggested many models of optical flow as how to explain accuracy of homing navigation of insects. In this paper, we introduce simple image matching based on optical flow to estimate the movement of agents for homing navigation. The optical flow made from Kanade-Lucas-Tomashi(KLT) algorithm has pattern regularity for agent movements. Hence, the agent can guess its own locomotion pattern, estimate distance from homimg direction, navigate through a distribution of patterns.