基本信息:
- 专利标题： Document processing apparatus and program
- 专利标题（中）：文件处理装置和程序
- 申请号：JP2008216184 申请日：2008-08-26
- 公开(公告)号：JP2010055142A 公开(公告)日：2010-03-11
- 发明人： ITONORI KATSUHIKO
- 申请人： Fuji Xerox Co Ltd , 富士ゼロックス株式会社
- 专利权人： Fuji Xerox Co Ltd,富士ゼロックス株式会社
- 当前专利权人： Fuji Xerox Co Ltd,富士ゼロックス株式会社
- 优先权： JP2008216184 2008-08-26
- 主分类号： G06K9/03
- IPC分类号： G06K9/03 ; G06F17/30 ; G06K9/00
摘要：
PROBLEM TO BE SOLVED: To obtain the same character recognition results from a character which should be recognized as one character before performing character recognition even when the distortion of a reading image is generated.
SOLUTION: A character image segmentation part 12 segments a character image from a document image input from an image input device 101, and a character image classification part 13 classifies the segmented character images. A mean character image feature acquisition part 15 generates image characteristics by averaging the classified character images for every category, and a character recognition part 16 performs character recognition to the averaged image characteristics. A recognition character code is uniformly assigned to the character images included in the category, and the character recognition results of a document are generated.
COPYRIGHT: (C)2010,JPO&INPIT
摘要（中）：
SOLUTION: A character image segmentation part 12 segments a character image from a document image input from an image input device 101, and a character image classification part 13 classifies the segmented character images. A mean character image feature acquisition part 15 generates image characteristics by averaging the classified character images for every category, and a character recognition part 16 performs character recognition to the averaged image characteristics. A recognition character code is uniformly assigned to the character images included in the category, and the character recognition results of a document are generated.
COPYRIGHT: (C)2010,JPO&INPIT
要解决的问题：即使在产生读取图像的失真时，也可以在执行字符识别之前从应该被识别为一个字符的字符获得相同的字符识别结果。 解决方案：字符图像分割部分12从从图像输入装置101输入的文档图像中分割字符图像，并且字符图像分类部分13对分割的字符图像进行分类。 平均字符图像特征获取部分15通过对每个类别的分类字符图像进行平均来生成图像特性，并且字符识别部分16对平均图像特征执行字符识别。 将识别字符代码均匀地分配给包括在类别中的字符图像，并且生成文档的字符识别结果。 版权所有（C）2010，JPO＆INPIT
公开/授权文献:
- JP4661921B2 Document processing apparatus and program 公开/授权日：2011-03-30
信息查询:
EspacenetIPC结构图谱:
| G | 物理 |
| --G06 | 计算；推算；计数 |
| ----G06K | 数据识别；数据表示；记录载体；记录载体的处理 |
| ------G06K9/00 | 用于阅读或识别印刷或书写字符或者用于识别图形，例如，指纹的方法或装置 |
| --------G06K9/03 | .错误的检测或校正，例如，用重复扫描图形的方法 |