Abstract
Stereo robot vision system achieve stereopsis by using two cameras. To solve the correspondence problem between right and left images, we proposed a pattern matching technology and applied the technology to a stereovision system. The proposed pattern matching technology employs a liquid crystal display (LCD) and utilizes the polarization of light. We use the LCD function of controlling light penetration for our pattern matching. We use two LCDs arranged in piles. The image from each camera is displayed on one of the LCDs; the correspondence between the two images is then judged by the light intensity through the two-layer LCD. Using this two-layer LCD, correspondence between complete objects in the images is roughly obtained, and the calculation for the pattern machining can be reduced. To confirm the effectiveness of the proposed technology, the stereo camera was mounted on a remote-control model car, and an experiment was conducted for automated driving. The images of the target vehicle and other objects in front of the autonomous vehicle obtained from the right and left cameras were displayed on the two-layer LCD, and the distances to the preceding car and other objects were calculated through triangulation.
