基于弹性笔画长度的汉字笔画平面抽取及识别

针对手写体汉字识别过程中的特征抽取，提出了一种改进的抽取笔画平面的方法。首先，将手写汉字图像进行非线性规一化；然后，利用弹性笔画长度，根据汉字的横竖撇捺四种笔画分别抽取出四种笔画的平面；最后，将四个笔画平面分别均匀划分成S×S个小网格，在每个小格内计算其笔画交叉数目，便得到了一个4S2维的特征向量。实验利用基于最小距离的分类器对含有7600（19类）个汉字的测试集进行了分类，取得了较好的识别效果。

Extraction of the stroke plane for the handwritten Chinese character recognition based on elastic stroke length and recognition

According to the feature extraction of the handwritten Chinese character recognition, an improved extracting stroke plane method was proposed. First, to nonlinearly normalize the Chinese character image; then, to extract the four stroke plane by making use of the elastic stroke length, according to the four strokes of heng, shu, pie, na; last, to cut each stroke plane into the S×S small meshing, and count the stroke crossing numbers in each meshing，then get a 4S2 character vector forms. The experiment adopted the minimal-distance classifier to classify the Chinese character set that includes 7600 characters, and achieved better recognition results.