特殊票据字符识别预分类特征的提取及分类字典的建立

CLASSIFICATION FEATURE EXTRACTION AND CLASSIFI-CATION LIBRARY SETTING UP OF SPECIAL TICKET'S OCR

从特殊票据———火车票票面字符的特点出发 ,将笔划复杂性指数与四周面积编码结合起来作为粗分类的分类特征 采用C -均值聚类算法进行预分类 最后生成分类特征库———分类字典 .得到了预期的分类效果 ,正确分类率达到 95 % .

Based on the special character of train tickets, the method using Index Complexity and Surrounding Area Coding as the classification feature is discussed.And by using c means algorithm to classify, the classification library is set up. The experimental result is satisfying. The accumulated classification rate reaches 95%.