首页> 外国专利> System for optical character recognition (ocr) using intelligent template matching.

System for optical character recognition (ocr) using intelligent template matching.

机译:使用智能模板匹配的光学字符识别(ocr)系统。

摘要

Template matching techniques are one of the simplest and oldest techniques in optical character recognition. The claimed invention uses template matching for character recognition. The invention does not require an entire input template to be matched but only the critical part of the input template. The invention does not require any mathematical formulae for selection of the template as well as for the actual comparison of templates. For any N x N size template a standard template matching needs N2 comparisons but the claimed invention requires (3n/2) comparisons at the maximum. The claimed system comprises of a counting block which determine the count of the input template which is nothing but a matrix with binary values. The system further includes classification block which classifies the given input. A target character repository provides a sequence of matching functions to select and recognize the correct character. The matching function block consists of all the matching functions that require simple binary comparison and by doing so avoid complicated mathematical computations. The performance of the preferred embodiment of the invention is measured by the number of comparisons required for each character input template for accurate selection and final recognition.
机译:模板匹配技术是光学字符识别中最简单,最古老的技术之一。所要求保护的发明使用模板匹配进行字符识别。本发明不需要匹配整个输入模板,而只需要匹配输入模板的关键部分。本发明不需要任何数学公式来选择模板以及模板的实际比较。对于任何N×N大小的模板,标准模板匹配需要N2个比较,但是所要求保护的发明最大需要(3n / 2)个比较。所要求保护的系统包括计数块,该计数块确定输入模板的计数,该输入模板不过是具有二进制值的矩阵。该系统还包括分类块,其对给定的输入进行分类。目标字符存储库提供一系列匹配功能,以选择和识别正确的字符。匹配功能块包含所有需要简单二进制比较的匹配功能,从而避免了复杂的数学计算。本发明的最佳实施例的性能是通过每个字符输入模板为精确选择和最终识别所需的比较次数来衡量的。

著录项

  • 公开/公告号IN2014MU00898A

    专利类型

  • 公开/公告日2015-10-30

    原文格式PDF

  • 申请/专利权人

    申请/专利号IN898/MUM/2014

  • 发明设计人 JOSHI SWANAND ARVIND;

    申请日2014-03-19

  • 分类号G06K9/00;G06F19/00;

  • 国家 IN

  • 入库时间 2022-08-21 15:14:35

相似文献

  • 专利
  • 外文文献
  • 中文文献
获取专利

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号