节点文献

基于Zernike矩的字符特征归一化方法

A method to normalize character feature based Zernike moment

  • 推荐 CAJ下载
  • PDF下载
  • 不支持迅雷等下载工具,请取消加速工具后下载。

【作者】 李了了邓善熙丁兴号

【Author】 (Dept. of Computer Information Science ,NeiJiang Teacher College, NeiJiang, SiChuan , China ,641112)Li Liaoliao (The School of Instrument Science and Opto- Electronic Engineering ,Hefei University of Technology,HeFei, AnHui , China 230009)DengShanxi (Dept. of Electronic Engineering, XiaMen University, XiaMen, FuJian , China, 361005) Ding Xinghao

【机构】 内江师范学院计算机信息科学系合肥工业大学仪器科学与光电工程学院厦门大学电子工程系

【摘要】 提取Zernike矩的幅值作为字符的模式特征具有旋转不变性,但单位圆半径的选择对特征计算影响较大。本文提出一种根据字符外部边框信息计算最佳单位圆半径的归一化方法。经实验证明该方法能保证单位圆合适地把整个字符包含在内,同时由于单位圆半径不变使数字图像离散化对Zernike矩旋转不变性的影响降到最低。

【Abstract】 The character’s pattern feature extracted from Zernike moment amplitude has rotation invariance, but the choice of unit circle radius affect the feature directly. The article presents a normalization method that can obtain the best unit circle radius based on information of character’s contour. Experiments demonstrated that the radius could cover the whole character properly by this method, at same time, because of radius invariance the influence of digital image’s discretization on rotation invariance of Zernike moment was reduced to the smallest level.

【关键词】 Zernike矩字符特征归一化
【Key words】 Zernike momentcharacter featurenormalization
【基金】 安徽省“十五”科技攻关重点资助项目(01022010)
  • 【分类号】TP391.41
  • 【被引频次】10
  • 【下载频次】305
节点文献中: 

本文链接的文献网络图示:

本文的引文网络