This paper describes a vision system that is for using character information for robots working on daily environments. It is mainly composed of image-based recognition module that detects character candidates by taking particular note of closed contour in an image. In addition, to cope with tiny or distant characters, a camera system with mechanical gaze adjustment is collaborated with the image processing. By combining the detection result with OCR (Optical Character Reader), applications to daily assistance by an autonomous robot were introduced.
展开▼
