公开(公告)号：US08560302B2

公开(公告)日：2013-10-15

申请号：US13127986

申请日：2009-02-25

申请人： Jinglian Gao ,  Binghui Chen

发明人： Jinglian Gao ,  Binghui Chen

IPC分类号： G06F17/27 ,  G06F17/20

CPC分类号： G06F17/276 ,  G06F3/0237 ,  G06F17/2755

摘要： The present invention provides a method for generating derivative words including the steps of: creating a number of derivative grammar arrays; matching the inputting character information with the derivative grammar arrays and obtaining the match derivative grammar arrays; obtaining match words from the language database according to the condition arrays of the obtained derivative grammar arrays and the inputting character information; and generating derivative words by adding the suffix alphabetic character sets of the obtained derivative grammar arrays to the ends of the words. In accordance with the established grammar rules, the words in the language database can be converted to derivative words and the derivative words do not need to be stored in the language database. Therefore, the storage space of the language database can be remarkably reduced. The present invention also provides a system for generating derivative words.

摘要翻译： 本发明提供一种产生导数词的方法，包括以下步骤：创建多个导数语法阵列; 将输入字符信息与导数语法数组进行匹配，得到匹配导数语法数组; 根据获得的导数语法阵列的条件数组和输入的字符信息，从语言数据库中获得匹配字; 并通过将获得的导数语法阵列的后缀字母集合添加到单词的末尾来生成导数词。 根据已建立的语法规则，语言数据库中的单词可以转换为派生词，并且派生词不需要存储在语言数据库中。 因此，可以显着地减少语言数据库的存储空间。 本发明还提供一种用于产生导数词的系统。

公开(公告)号：US08396295B2

公开(公告)日：2013-03-12

申请号：US13142856

申请日：2009-02-25

申请人： Jing-lian Gao ,  Xinchun Huang ,  Binghui Chen ,  Anjin Hu ,  Muyu Cai ,  Huaxing Lu ,  Zhipin Liu ,  Zhiai Wang ,  Fang Guo ,  Jingping Li ,  Honghui Wang ,  Chuntao Tan ,  Zhengwei Wu

发明人： Jing-lian Gao ,  Xinchun Huang ,  Binghui Chen ,  Anjin Hu ,  Muyu Cai ,  Huaxing Lu ,  Zhipin Liu ,  Zhiai Wang ,  Fang Guo ,  Jingping Li ,  Honghui Wang ,  Chuntao Tan ,  Zhengwei Wu

IPC分类号： G06K9/00

CPC分类号： G06K9/00416 ,  G06K9/00422 ,  G06K9/6234

摘要： The present invention discloses a method for recognizing a handwritten character, which includes the following steps of: obtaining a coarse classification template and a fine classification template; receiving a handwritten character input signal from a user, gathering a discrete coordinate sequence of trajectory points of the inputted character, and pre-processing the discrete coordinate sequence; extracting eigenvalues and calculating a multi-dimensional eigenvector of the inputted character; matching the inputted character with the coarse classification template to select a plurality of the most similar candidate character classes; and matching the eigen-transformed inputted character with sample centers of the candidate character classes selected from the fine classification template, and determining the most similar character classes among the candidate character classes. The present invention further discloses a system for recognizing a handwritten character. The present invention can recognize an inputted character fast at a high recognition precision.

摘要翻译： 本发明公开了一种用于识别手写字符的方法，包括以下步骤：获得粗分类模板和精细分类模板; 从用户接收手写字符输入信号，收集输入字符的轨迹点的离散坐标序列，并预处理离散坐标序列; 提取特征值并计算输入字符的多维特征向量; 将输入的字符与粗分类模板匹配以选择多个最相似的候选字符类; 以及将所述本征变换输入字符与从所述精细分类模板中选择的候选字符类别的样本中心进行匹配，以及确定所述候选字符类别中最相似的字符类别。 本发明还公开了一种用于识别手写字符的系统。 本发明能够以高识别精度快速识别输入的字符。

公开(公告)号：US20110311141A1

公开(公告)日：2011-12-22

申请号：US13142856

申请日：2009-02-25

申请人： Jinglian Gao ,  Xinchun Huang ,  Binghui Chen ,  Anjin Hu ,  Muyu Cai ,  Huaxing Lu ,  Zhipin Liu ,  Zhiai Wang ,  Fang Guo ,  Jingping Li ,  Honghui Wang ,  Chuntao Tan ,  Zhengwei Wu

发明人： Jinglian Gao ,  Xinchun Huang ,  Binghui Chen ,  Anjin Hu ,  Muyu Cai ,  Huaxing Lu ,  Zhipin Liu ,  Zhiai Wang ,  Fang Guo ,  Jingping Li ,  Honghui Wang ,  Chuntao Tan ,  Zhengwei Wu

IPC分类号： G06K9/18

CPC分类号： G06K9/00416 ,  G06K9/00422 ,  G06K9/6234

摘要： The present invention discloses a method for recognizing a handwritten character, which includes the following steps of: obtaining a coarse classification template and a fine classification template; receiving a handwritten character input signal from a user, gathering a discrete coordinate sequence of trajectory points of the inputted character, and pre-processing the discrete coordinate sequence; extracting eigenvalues and calculating a multi-dimensional eigenvector of the inputted character; matching the inputted character with the coarse classification template to select a plurality of the most similar candidate character classes; and matching the eigen-transformed inputted character with sample centers of the candidate character classes selected from the fine classification template, and determining the most similar character classes among the candidate character classes. The present invention further discloses a system for recognizing a handwritten character. The present invention can recognize an inputted character fast at a high recognition precision.

摘要翻译： 本发明公开了一种用于识别手写字符的方法，包括以下步骤：获得粗分类模板和精细分类模板; 从用户接收手写字符输入信号，收集输入字符的轨迹点的离散坐标序列，并预处理离散坐标序列; 提取特征值并计算输入字符的多维特征向量; 将输入的字符与粗分类模板匹配以选择多个最相似的候选字符类; 以及将所述本征变换输入字符与从所述精细分类模板中选择的候选字符类别的样本中心进行匹配，以及确定所述候选字符类别中最相似的字符类别。 本发明还公开了一种用于识别手写字符的系统。 本发明能够以高识别精度快速识别输入的字符。