A 3-layer recurrent neural network is proposed for parallel thinning. This network iteratively removes the contour points of an object shape by template matching. After the neural network converges, a perfectly 8-connected skeleton is derived. The set of templates which the neural network tries to match is specially designed for a one-pass parllel thinning algorithm. The proposed neural network is shown both theoretically and experimentally to do the same work as the one-pass parallel thinning algorithm.

Keywords: parallel thinning, recurrent neural network, template matching, excessive erosion

Received May 31, 1993; revised November 23, 1993.
Communicated by Soo-Chang Pei.
^*This work was supported by the National Science Council, Republic of China, under grant No. NSC81-0404-E009-017.