质心层次特征的无约束手写体数字识别

光学字符识别（ＯＣＲ）是模式识别最为成功的应用之一．目前，ＯＣＲ的研究重点是无约束手写体字符识别．采用了基于字符质心的层次特征对无约束手写体数字进行分类识别．基于字符质心的不均匀分块方法，在一定程度上可以克服无约束手写体数字字形千变万化所引起的不稳定性．层次特征将字符在空间的二维分布转化为一维，特征抽取过程简单，易于实现．将该算法应用于无约束手写体数字的信函分拣系统，单字的平均识别率达９７％以上．

Recognition of Unconstrained Handwritten Numerals Based on Centroid Layer Feature

Optical character recognition (OCR) is one of the most successful applications of pattern recognition. At present, an emphasis of the research is put on unconstrained handwritten character recognition. An algorithm to recognize unconstrained handwritten numerals based on centroid layer feature is proposed in this paper. The unsymmetrical separating method based on centroid is stable to a certain degree. The layer feature transforms 2 dimension distribution of the character into 1 dimension, and the extraction of this feature is easy to realize. The experiments show that the average recognition rate of single is above 97%.