Abstract: In order to further enhance the effect of multi-focus image fusion, a coherence based algorithm is presented. With the high-frequency coefficients after the discrete wavelet transform, a structure tensor matrix is calculated. The coherence defined by the eigenvalues of the structure tensor conveys the local geometrical features and is used for device a new fusion rule. The experimental results on test images demonstrate the performance of the proposed method quantitatively and visually.

Key words: multi-focus image fusion, wavelet transform, structure tensor, eigenvalues, coherence

摘要： 为提高多聚焦图像的融合效果，提出一种基于相干性的融合算法。该算法对源图像进行离散小波变换，利用高频小波系数构造结构张量矩阵，通过矩阵特征值定义反映局部几何信息的相干性并建立融合策略。实验结果表明，该算法得到的融合图像在主观视觉效果和客观量化指标方面均有良好的表现，提高了融合的视觉效果。

关键词: 多聚焦图像融合, 小波变换, 结构张量, 特征值, 相干性