【摘要】 针对目前常见的脱机手写数字识别技术进行分析比较,并模拟人工智能识别字符的过程,提出一种分层识别方法。对于结构差异较大的数字字符,根据数字字符图像的统计特征进行粗略识别;然后进一步利用结构特征细致区分形状相近的数字字符。通过自制样本库进行识别测试,结果表明该方法能够有效提高识别准确率。更多还原

【Abstract】 Analyzing and comparing the common off-line handwritten numeral recognition methods,and simulating the produce of character recognition of human,a hierarchical recognition method was proposed. Firstly,the principal component analysis method was adopted to extract statistical character of numeral image. In order to further determine the type of numeral,the structure character of width and height rate should be added. By self-regulating numeral samples,experiments were implemented,the accuracy of this method of digital recognition was verified.更多还原