Design and recognition of a ring code for AGV localization

The key issue in ring code detection is to locate the center of the rings. The Hough transform can perform this task well. However, it is slow for real time applications. A novel algorithm is proposed in combination with one-pass chain-generating algorithm. In this algorithm, two random points are selected on a single chain. If these two points are on the edge of the same circle, a perpendicular line passing through the center of this segment will also pass through the center of the circle. Intersection of two such lines will locate the center of the circle. The image space is divided into equally spaced bins to collect intersection points. We are looking for only one center of the ring code. Therefore, one bin with the largest collection of intersection points will be identified. This algorithm has the added advantage of very low computational cost comparing with the traditional one. The experimental results show that the ring code can be effectively recognized even under very noisy environment and with partially covered ring code.