A run-length-coding-based approach to stroke extraction of Chinese characters

Traditional stroke extraction approach usually adopts thinning technique as the preprocessing method in obtaining the skeletons of Chinese characters. However, thinning may produce spurious branches and multiple fork points a t junctions. Such distortion will make stroke extraction process more compl icate and unreliable. This paper proposes a novel run-length-based stroke e xtraction approach without using the thinning method. Besides, the proposed approach does not need to trace the skeleton pixel by pixel in obtaining t he skeletons of Chinese characters. In our approach, run-length coding tech nique is first employed to get a special skeleton which only owns disjoint line segments without including fork points. Then, an attributed graph is c onstructed from the skeleton. The attribute between two nodes is determined according to the distance, connectivity and orientation difference between the two corresponding line segments. Intersection relation among line segm ents is represented by a junction matrix and its associating graph. While s troke extraction is performed, fork points can also be found. Experimental results show that the proposed approach is feasible and efficient in extrac ting strokes of Chinese characters. (C) 2000 Pattern Recognition Society. P ublished by Elsevier Science Ltd. All rights reserved.