【摘要】 在用句法结构识别限制性手写汉字系统中,利用笔划作为基元,由机器自动地抽取笔划是困难的。 本文提出了一种从汉字骨架中抽取笔划的新方法:用启发式搜索和超前搜索抽取笔划成分,再分析与合成得到笔划,在抽取笔划成分时,考虑到细化引起的畸变。文中给出了搜索笔划成分的框图,对922个手写汉字进行了模拟实验,抽取笔划的正确率达到94％。更多还原

【Abstract】 In handprinted Chinese character recognition systems based on syntactic structures, it is difficult to extract strokes automatically with stroke elements. In this paper, a new method for extracting strokes from skeletons of Chinese characters is presented. With the method, components of strokes are extracted by means of heuristic search and forward search, and then analyzed and composed. The distortion caused by thinning is also considered. Simulation results show that the correct recognizing rate is 94 percent.更多还原