Computer Engineering and Applications ›› 2008, Vol. 44 ›› Issue (19): 190-192.

• 图形、图像、模式识别 • Previous Articles     Next Articles

Frequency domain approach based on mutual information to image registration

WU Ya-juan,LIU Xiao-feng   

  1. College of Computer,China West Normal University,Nanchong,Sichuan 637002,China
  • Received:2008-02-27 Revised:2008-05-05 Online:2008-07-01 Published:2008-07-01
  • Contact: WU Ya-juan

一种基于互信息的图像频域配准算法

吴亚娟,刘晓锋   

  1. 西华师范大学 计算机学院,四川 南充 637002
  • 通讯作者: 吴亚娟

Abstract: Image registration is the key step of image analysis with images from multiple resources.It is also the foundation of image application.The advantages of frequency approaches to registration are rapidness and high precision.The frequency approach proposed by P.Vandewalle outperforms previous frequency methods and some special domain methods.In this paper,it is improved to be faster and get higher precision.Finally,a frequency domain approach based on mutual information to image registration is presented by using half the amplitude of the Fourier transform and mutual information theory.

Key words: registration, cross-power spectrum, mutual information, amplitude of the Fourier transform

摘要: 图像配准是多源图像分析的关键步骤,是图像应用的基础。频域配准方法具有配准精度高和速度快的优点。P.Vandewalle的频域配准算法明显优于其他频域算法和一些空间域算法,对该算法进行了改进,仅使用了一半图像频谱灰度,在对分块后的频谱灰度进行分析时引入了互信息理论,实现了配准精度更高、速度更快的基于互信息的图像频域配准算法。

关键词: 配准, 互功率谱, 互信息, 傅里叶频谱