计算机与现代化

• 图像处理 • 上一篇    下一篇

基于四元数傅里叶梅林变换的旋转不变彩色纹理分类

  

  1. 华南师范大学物理与电信工程学院,广东广州510006
  • 收稿日期:2014-02-26 出版日期:2014-05-28 发布日期:2014-05-30
  • 作者简介:武静(1988-),女,河南南阳人,华南师范大学物理与电信工程学院硕士研究生,研究方向:数字图像处理与模式识别。

Rotation-invariant Color Texture Classification Based on Quaternion Fourier-Mellin Transform

  1. School of Physics and Telecommunication Engineering, South China Normal University, Guangzhou 510006, China
  • Received:2014-02-26 Online:2014-05-28 Published:2014-05-30

摘要: 提出一种基于四元数傅里叶梅林变换(Quaternion Fourier-Mellin Transform,QFMT)的旋转不变彩色纹理分类方法。该方法首先对彩色图像各分量图像进行对数极坐标变换,然后将经过变换后的3幅分量图像表示成四元数,并对其进行四元数傅里叶变换(Quaternion Fourier Transform,QFT),最后对幅度谱分别统计其环形特征量和楔形特征量作为纹理分类的特征向量,利用最近邻分类器进行分类。实验结果表明,本文提出的方法分类准确率更高,且具有良好的旋转不变纹理分析性能。

关键词: 四元数傅里叶梅林变换, 对数极坐标, 旋转不变, 彩色纹理分类

Abstract: A method of color texture classification based on quaternion Fourier-Mellin transform is proposed, which is rotation-invariant. Firstly, log-polar coordinate transform is used on each component image of the color image, and then the three components which have been transformed are represented as a quaternion and carries on the quaternion Fourier transform; finally, statistic of the ring and wedge characteristic feature of the amplitude spectrum acts as texture classification vector, and uses the nearest neighbor classifier to classify. The experimental results show that the classification accuracy of the proposed method is higher and is of a good rotation invariance.

Key words: quaternion Fourier-Mellin transform, logarithmic polar coordinates, rotation invariant, color texture classification

中图分类号: