计算机工程与应用 ›› 2012, Vol. 48 ›› Issue (20): 182-186.

• 图形、图像、模式识别 • 上一篇    下一篇

一种改进和提取Zernike特征矩的方法

陈  玲1,3,李熙莹2,3,罗  玲2,3   

  1. 1.中山大学 物理科学与技术学院,广州 510275
    2.中山大学 工学院,广州 510275
    3.广东省智能交通系统重点实验室,广州 510275
  • 出版日期:2012-07-11 发布日期:2012-07-10

Method of modification and feature selection of Zernike moment

CHEN Ling1,3, LI Xiying2,3, LUO Ling2,3   

  1. 1.School of Physics and Engineering, Sun Yat-Sen University, Guangzhou 510275, China
    2.School of Engineering,Sun Yat-Sen University,Guangzhou 510275, China
    3.Guangdong Provincial Key Laboratory of Intelligent Transportation System, Guangzhou 510275, China
  • Online:2012-07-11 Published:2012-07-10

摘要: 数字图像的离散性使Zernike矩的计算存在误差,降低了特征描述能力。采用一种方圆转换方法将矩形图像转化为抽象的圆形图像,计算转化后图像Zernike矩,改进Zernike矩的计算方法,从而减少计算误差,提高计算速率,提高Zernike矩的特征描述能力。同时根据统计理论,提出一种规范的特征提取方法。经车牌字符识别实验验证,改进方法能够很好地提取图像特征,用于模式识别。

关键词: Zernike矩, 方圆变换, 特征选择

Abstract: The accuracy of Zernike moment is a significant problem in pattern recognition due to the discrimination of image. By transforming the square image to an abstract circle image using a square-to-circle method, the computation of Zernike moment is speeded up, and the representation ability of Zernike moment is improved. Combining with statistical feature selection theory, suitable eigenvector construction method is proposed. The result of plate license character recognition shows this method can work efficiently.

Key words: Zernike moment, square to circle, feature selection