计算机工程

• 图形图像处理 • 上一篇    下一篇

基于不可分小波分解的图像配准方法

刘 斌,孙 斌,余方超,唐虎潇   

  1. (湖北大学数学与计算机科学学院,武汉430062)
  • 收稿日期:2013-09-30 出版日期:2014-10-15 发布日期:2014-10-13
  • 作者简介:刘 斌(1963 - ),男,教授、博士、博士生导师,主研方向:图像处理,小波分析与应用;孙 斌,讲师、硕士;余方超,硕士研究 生;唐虎潇,本科生。
  • 基金项目:
    国家自然科学基金资助项目(61072126);湖北省自然科学基金资助重点项目(2012FFA053)。

Image Registration Method Based on Nonseparable Wavelet Decomposition

LIU Bin,SUN Bin,YU Fang-chao,TANG Hu-xiao   

  1. (School of Mathematics and Computer Science,Hubei University,Wuhan 430062,China)
  • Received:2013-09-30 Online:2014-10-15 Published:2014-10-13

摘要: 张量积小波强调的是图像中水平和垂直方向的高频信息,而不可分小波具有各向同性,可以提取图像中各个方向的边缘,能获得比较完整的图像轮廓,将这种特点应用于图像配准时,能准确定位图像仿射不变点的位置。为此,提出一种通过求取不可分小波分解后的高频子图像配准参数来配准原图像的方法,把图像的配准问题转化为其不可分小波分解后的高频子图像配准问题。从不可分小波分解的快速算法理论出发,证明该配准方法的正确性。构造一组四通道不可分小波滤波器组,在此基础上给出配准的方法和步骤。实验结果表明,该方法具有较好的配准效果,其求取图像配准参数的运算量比直接求取原图像配准参数运算量的1/ 4 还少,与基于张量积小波分解的图像配准方法相比,具有较高的配准精度。

关键词: 图像配准, 不可分小波, 仿射变换, 滤波器组, 质心点, 加权质心点

Abstract: Tensor product wavelet only emphasizes on the edge of the horizontal and vertical direction. Nonseparable wavelet is isotropic,can extract the edge of the image in all directions,and can obtain relatively complete outline of the image. When this kind of characteristics is applied to the image registration,it can acquire the accurate positions of the invariant points of the affine transform. Based on this characteristic,this paper proposes an image registration method through calculating the registration parameters of the high-frequency sub-images of the nonseparable wavelet decomposition of the original image. This method can transform the registration of the original image into the registration of its high-frequency sub-images. The correctness of the proposed registration method is proved according to the fast algorithm theory of wavelet decomposition. A four channel nonseparable wavelet filter bank is constructed and the registration steps of the method are given. Experimental results show that the method has better registration results. The amount of computation for computing the registration parameter is less than a quarter of the registration parameters computation of the original image. When compared with registration method based on the tensor product wavelet decomposition,the method has higher registration precision.

Key words: image registration, nonseparable wavelet, affine transform, filter bank, centroid point, weighted centroid point

中图分类号: