摘要:
本文介绍用于联机手写中文字自动识别的新方法与新算法.由于下述各点的实现,手写
文字时可以减少许多限制,增加书写自由.①笔划的抽取经由两次分段实现:首先连续采样,
将输入笔划变换成线段组成,再对线段的长度进行比较,删去相对不重要的成份.②用笔划校
正技术将不应分离的笔划重新组合成规范笔划,或者将不应联写的复合笔划重新分解成基本
笔划.③用非完全匹配技术使失真字可以识别.④用混序笔划重排算法可使一个混序笔划输
入的字重新排列笔顺.⑤笔划位置和长度作为进一步特征,可区别模糊字.
Abstract:
In this paper, a few new techniques and algorithms for computer on-line handwritten
Chinese chiaracter recognition are presented. More freedom in writing can be provided by the
following points:(1) sementation is based firstly on successive sampling and secondly on a
method which eliminates, based on a local comparison, the undesired relative short segments of
a strok;(2) by means of correction algorithm, strokes which should not be separated can be linked
together, and strokes which should not be linked can be split into several basic strokes;(3)
the distorted characters can be recognized by a non-exact matching process;(4) with reference
to the stroke sequence order of a template character stored in the dictionary, the stroke sequence
order of an input character can be rearranged;(5) use stroke position and stroke length as
additional features to distinguish ambiguous characters.