Detection of Divergence Point of the Optical Flow Vectors Considering to Gaze Point While Vehicle Cornering

This paper discusses on the relationship between a gaze direction and an optical flow and proposes a divergence point detection from optical flow vectors in a cornering behavior. Previously, we express optical flow radiates outward from the gaze center using the movie of the real vehicle. However, it is difficult to detect a divergence point corresponding with gaze position by applying vector analysis, because the acquired optical flow vectors considering with gaze position includes blank area and noise by matching error. Therefore, we propose the divergence point detection method based on particle swarm optimization. For verifying the proposed method, we perform the experiment using real driving movie. We discuss the effectivity and issue of the proposed method.