Abstract:  For improving the performance of multi-focus image fusion and solving the problem that parameter estimation is short of accuracy in the statistical image fusion method, a novel multi-focus image fusion approach is proposed based on Principal Component Analysis（PCA） and generalized Gaussian modeling in wavelet transform domain. First, coefficients of the high frequency subbands are modeled as generalized Gaussian distribution, and the fusion parameters are estimated by an improved maximum likelihood method. Then, coefficients of the low frequency subbands are fused by a method of regional PCA. And finally, an effective fused image is achieved through inverse wavelet transform. The experimental results demonstrate that, comparing with existing multi-focus image fusion approaches, the proposed method makes the fused image have more information and better visual effect.

Key words: wavelet transform, generalized Gaussian modeling, principal component analysis, parameter estimation, multi-
focus image fusion

摘要： 为了提高多聚焦图像的融合性能，针对统计图像融合方法中像素间融合参数估计的不足，提出一种基于小波变换域主成分分析（PCA）和广义高斯建模的多聚焦图像融合方法。将高频子带系数建模为广义高斯分布，并通过改进的最大似然估计法获取融合参数。结合低频子带系数的区域PCA融合方法，最终实现有效的图像融合。实验结果表明，该方法与传统的多聚焦图像融合方法相比，可使融合图像的信息量更丰富，具有更佳的视觉效果。

关键词: 小波变换, 广义高斯建模, 主成分分析, 参数估计, 多聚焦图像融合