Abstract: Aiming at the problems of edge blurring and image distortion of traditional wavelet transform in image fusion, a multi-focus image fusion algorithm based on super-resolution is proposed in this paper. All source images are processed by double cubic interpolation in single frame super-resolution to enhance the source image contrast and other details. The source images used in this paper are two images in the same scene which are processed by left and right focusing respectively. Stationary Wavelet Transform（SWT） is implemented for these high resolution source images, and the source images are divided into four sub-bands. To solve the problems of source image detail information confusion and structural information redundancy contained in these subbands, Principal Component Analysis（PCA） is adopted to select the maximum signal-to-noise ratio of each subband of the source image for image fusion. Inverse Stationary Wavelet Transform（ISWT） is used to reconstruct the fusion subband and obtain high quality fusion images. In order to evaluate the quality of fused image, two measurement methods, no reference image and full reference image, are selected to detect the quality of fused image. The experimental results show that the proposed algorithm overcomes the shortcomings of traditional wavelet transform algorithm in image fusion, and has the advantages of clear edge, good visual perception, good clarity and low distortion.

Key words: multi-focus image fusion, super resolution, stationary wavelet transform

摘要： 针对传统小波变换在图像融合过程中出现边缘模糊、图像失真等问题，提出了一种基于超分辨率的多聚焦图像融合算法。对所有的源图像进行了双三次插值的单帧超分辨率处理，增强源图像对比度等细节信息，采用的源图像为分别进行左右聚焦处理的同一场景中的两幅图像。对这些高分辨率源图像实现了平稳小波变换（SWT），并将源图像划分为四个子带。针对这些子带所包含源图像细节信息混乱、结构信息冗余等问题，采用了主成分分析（PCA），分别选取源图像各子带的最大信噪比进行图像融合。利用逆平稳小波变换（ISWT）对融合子带进行重构，得到高质量融合图像。为了评定融合后图像的质量，选择了无参考图像和全参考图像的两种度量方法来检测融合后的图像质量。经实验结果表明，提出的算法克服了传统小波变换算法在图像融合上的缺点，具有边缘清晰、视觉感知好、清晰度好、失真小等优点。

关键词: 多聚焦图像融合, 超分辨率, 平稳小波变换