打印文件鉴别打印机型的文字图像模糊识别*

探索了一种以打印件鉴别打印机型的文字图像计算机模糊识别方法.该方法收集标准常用字号和字体,以及常用打印机打印的文字,扫描采集,用改进的直方图波形分析法处理图像,提取文字的笔画总面积和笔画轮廓总周长等特征指标;再选定一种机型为参照,对各种机型相同字上述指标测量值及其几种组合的计算值,形成相对差值指标序列,建立信息数据库.在此基础上,建立对应指标的统计均值波动区间的值域表,并确定各指标的权重和建立权重系数矩阵.判断未知机型时,先按照前述方法任测100个常用字,利用OCR汉字识别模块和前述指标,自动辨识文字,进入模糊识别过程.根据相应检测字在值域表区间出现的概率,建立模糊关系矩阵.通过两个矩阵乘积的模糊变换产生判别矩阵.以最大隶属性确定打印机类型.按照数学模型,设计并实现打印机智能鉴别程序.应用实例测试,结果显示判别准确,符合设计预期.

Printer identification based on computer fuzzy recognition of character image

A new method about computer fuzzy recognition of character image was proposed to identify the type of printer. First, collected and disposed common and standard characters in main fonts which eigenvalues were extracted in area and perimeter of characters. Second, as a reference, the eigenvalues of one printer were extracted to make the sequence of relative difference of eigenvalue and to build database. Third, made the table of statistical mean eigenvalue in fluctuant range and built the matrix of weighted eigenvalues. To identify unknown printer, the eigenvalues of common 100 characters were measured in above method. Then, OCR component was used to recognize characters automatically and to search relevant database rapidly. It could get the identification matrix by fuzzy transform of eigenvalues matrix and weighted coefficient matrix that would be got by probability of recognized characters in eigenvalue table. The type of printer could be identified on fuzzy membership principle. The experimental result shows that identification program on above mathematic model is accordant with the anticipated design.: printer; identification; character image; fuzzy recognition