A Chinese-Character Stroke Extaction Algorithm Based on Contour Information

Abstract

In this paper, we present a new stroke extraction algorithm that integrates all levels of contour information including boundary points, dominant points, corner points, segments, cross section sequence graph, character structure to extract stokes of Chinese characters. Unlike the traditional bottom-up approach, which the current step of information extraction is only based on the previous step, we use and accumulate all extracted information until the final extraction of strokes is done. The longer contour segment of stroke tends to preserve better stroke information and is robust to the interfering contour. Hence, stoke extraction in the proposed algorithm is based on the longer contour segment of stroke instead of the medial axis. Experimental results show that the proposed algorithm can correctly extract the strokes up to 95% from a printed and handwritten test samples based on the human perception. This research provides a solid basis for the structural matching of Chinese characters.