作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程 ›› 2006, Vol. 32 ›› Issue (6): 215-217.

• 人工智能及识别技术 • 上一篇    下一篇

一种新的表格识别特征提取方法

黄锦德,郝红卫,张冬霞   

  1. 北京科技大学信息工程学院,北京100083
  • 出版日期:2006-03-20 发布日期:2006-03-20

A New Feature Extraction of Forms Recognition

HUANG Jinde, HAO Hongwei, ZHANG Dongxia   

  1. School of Information Engineering, Beijing University of Science and Technology, Beijing 100083
  • Online:2006-03-20 Published:2006-03-20

摘要: 采用一种表格识别方法实现对多种类型表格的识别,系统利用表格投影轮廓的功率谱密度作为表格的不变性特征向量。为了解决具有相互对称结构表格的识别问题,提出一种新的特征提取方法:采用区域划分的策略,综合考虑表格图像在水平方向及垂直方向上的特征,以分区投影轮廓的功率谱密度作为表格图像的特征向量。实验表明,这种方法能够有效解决具有对称结构表格的识别问题。

关键词: 特征提取;区域划分;表格识别;功率谱密度

Abstract: A method of forms identification is introduced, which can be employed in processing of documents with a certain style of application form. In order to solve the identification of such forms with a symmetry figure, a new approach of feature extraction is presented, which is a method of area division done by partition the forms image into areas. Both the horizontal feature and the vertical feature of forms image are token into account, and the power spectral density of the subareas’ profile is used as forms image’s feature vector. It is shown to be an effective method to identify the forms with symmetry figure

Key words: Feature extraction; Area division; Form identification; Power spectral density