计算机应用 ›› 2012, Vol. 32 ›› Issue (05): 1283-1285.

• 图形图像处理 • 上一篇    下一篇

基于具有对称性不可分小波的多聚焦图像融合

李凯,刘斌   

  1. 湖北大学 数学与计算机科学学院,武汉 430062
  • 收稿日期:2011-09-18 修回日期:2011-12-08 发布日期:2012-05-01 出版日期:2012-05-01
  • 通讯作者: 李凯
  • 作者简介:李凯(1987-),女,湖北荆州人,硕士研究生,CCF会员,主要研究方向:小波理论与应用、图像融合;刘斌(1963-),男,湖北黄冈人,教授,博士生导师,主要研究方向:小波理论与应用、图像融合、模式识别、图像配准。
  • 基金资助:

    国家自然科学基金资助项目(61072126);湖北省自然科学基金重点项目(2009CDA133)

Multi-focus image fusion based on non-separable symmetric wavelets

LI Kai,LIU Bin   

  1. College of Mathematics and Computer Science, Hubei University, Wuhan Hubei 430062, China
  • Received:2011-09-18 Revised:2011-12-08 Online:2012-05-01 Published:2012-05-01
  • Contact: LI Kai
  • Supported by:

    ;the Natural Science Foundation of Hubei province of China

摘要: 针对可分小波多聚焦图像融合方法存在的不足,提出一种基于四通道不可分小波的多聚焦图像融合方法。首先根据不可分小波理论,构造出一组二维四通道4×4具有对称性的不可分小波滤波器组;然后利用此滤波器组对参加融合的图像进行滤波,低频部分采用简单的加权平均算法,高频部分采用局部窗口能量取大的融合算法对分解后的系数图像进行融合;最后对图像进行重构,并采用熵、平均梯度等指标对融合结果图像进行了评价。实验结果表明,该方法对多聚焦图像的融合有较好的融合效果,与采用相同融合算法的基于可分小波的融合方法相比,能更好地突出低频域边缘细节信息,得到更为清晰的融合结果图像。

关键词: 图像处理, 不可分小波, 图像融合, 滤波器组

Abstract: A new fusion method of multi-focus images based on the four-channel non-separable wavelet was proposed, which aimed to solve the problem which exists in the separable wavelet-based fusion methods. First, a 4×4 non-separable wavelet 4-channel filter bank with linear phase using the theory of non-separable wavelets was constructed. Then images involving the fusion were decomposed by using the filter bank, for low-frequency part, the average value was selected, for the three high-frequency parts of each level, the value of the area window whose energy was bigger was selected. Finally, the new fused image was reconstructed. The performance of the method was evaluated using entropy, average gradient, etc. The experimental results show that it has good effect on the fusion of multi-focus images. The performance is better than that of the separable wavelet fusion method by using the same fusion algorithm. According to this method, the fused images are clearer and the detailed edge information of low-frequency domain is better obtained.

Key words: image processing, non-separable wavelet, image fusion, filter bank

中图分类号: