印刷体数字识别研究与实现

通过对数字识别算法的研究和印刷体数字特征进行分析，提出一种通过统计预处理后的已知类别样本的特征点分布概率为模板，并通过计算待识别样本与模板距离来确定待识别样本所属类别的方法。该方法运行速度快，识别精度高，但只局限于对印刷体数字的识别。实验结果显示，该方法对印刷体数字的识别率可达99%左右，优于一般结构特征识别方法。

Research and Implementation of Digital Recognition of Printed Number

Through the research of digit recognition algorithm and analysis on characteristics of printed digital, it pro-posed a method that through the statistics of probability distribution of the feature points of the known samples as a template,and calculation the distance from unknown samples and template to determine the category of unknown sam-ples.The method is fast and accurate,but it is only limited to the recognition of the printed body.The experimental re-sults show that the recognition rate of the printed digital reach 99%. The method is better than the general structural feature recognition method for recognition rate.