摘要： 传统二维经验模式分解图像融合方法以像素点能量最大原则作为融合依据，而不分析图像的特征信息，特征信息得不到最大保留。为此，提出基于二维经验模式分解与非负矩阵分解的图像融合方法。通过二维经验模式分解得到图像的内蕴模式函数和剩余量，并对内蕴模式函数进行非负矩阵分解，提取真实内蕴模式函数作为图像融合后的内蕴模式函数，利用反向重构得到融合图像。实验结果表明，该方法在图像清晰度与对比度方面均优于二维经验模式分解与非负矩阵分解方法。

关键词: 二维经验模式分解, 非负矩阵分解, 多尺度分解, 图像融合, 图像增强, 多源遥感图像

Abstract: The image fusion of traditional two-dimensional empirical mode decomposition is based on the fusion principle of maximum energy to pixels, but it does not analyze the characteristics information of images, which can not reserved maximum. Aiming at this problem, an image fusion method is proposed based on Bidimentional Empirical Mode Decomposition(BEMD) and Non-negative Matrix Factorization(NMF). Original images are decomposed using BEMD separately and the Intrinsic Mode Functions(IMF) component and residual component are obtained. Then the same level IMF of original images is decomposed using NMF separately and the real IMF as this level image fusion is obtained. The fussed image is obtained using reverse reconstruction. Experimental results show that this method is better than BEMD and NMF on the image clarity and contrast.

Key words: Bidimentional Empirical Mode Decomposition(BEMD), Non-negative Matrix Factorization(NMF), multi-resolution decomposition, image fusion, image enhancement, multi-source remote sensing image

中图分类号: 

• TN911.73