结合向导滤波与复轮廓波变换的多聚焦图像融合算法

多聚焦融合在图像识别和分析中具有非常重要的地位。为了有效地保留源图像的细节，克服变换域算法由于空间不连续性产生的人造纹理和灰度不均衡。该文结合复轮廓波时频分离的优点和向导滤波的特点提出了一种基于向导滤波的复轮廓波域多聚焦图像融合算法。首先，对源图像进行复轮廓波分解，其次对分解的低频系数进行基于向导滤波的均值融合策略，然后对分解的高频系数进行基于向导滤波的改进拉普拉斯能量和模取大的融合策略，最后通过复轮廓波反变换得到融合后的图像。实验结果表明，该算法利用向导滤波显著的提升了变换域融合算法的空间连续性，不仅可以获得良好的视觉融合效果，而且其客观评价指标也得到了提升。 

Multi-focus image fusion using guided filtering and complex contourlet transform

Multi-focus image fusion plays important roles in image recognition and analysis. In order to preserve the details of source images well and overcome artificial textures and non-uniform grayscale that is produced by lacking spatial continuity in transform domain fusion algorithms. A new multi-focus image fusion algorithm based on guided filtering via complex Contourlet domain is proposed. The new method holds the advantage of time-frequency separation of complex Contourlet and the feature of guided filtering. First, complex Contourlet is utilized for decomposition of the source images, and then apply average weight fusion rule with guided filtering to the low frequency coefficients of complex Contourlet, and apply larger sum-modified-Laplacian with guided filtering fusion rule to the high frequency coefficients of decomposition, finally the fusion image is reconstructed by the inverse complex Contourlet. The algorithm uses the guided filtering significantly enhance the transform domain fusion algorithms’ spatial continuity. Experimental results demonstrate that the proposed fusion method can both obtain good visual effect and highly objective.