基于多维自适应滤波与泊松融合的多聚焦图像融合算法

针对现有的多聚焦图像融合算法，在抗噪性能、图像连续性以及时间复杂度上的不足，提出了一种基于多维自适应滤波与泊松融合的多聚焦图像融合算法。首先，对统一尺度的源图像构建梯度矩阵，并设计尺度分段的多维自适应核函数对矩阵进行分割；然后，引入判定条件，对泊松重建中的图像边界进行伪边缘抑制，得到纯化后的全聚焦图像；最后，将抗噪算法与融合算法分别对比其他算法模型，进行综合比较，通过6种图像质量评价指标来评估融合性能与效果。结果表明：本文算法不仅有良好的运行效率，清晰的视觉观感，而且在评价因子上有更理想的数据结果。可见，不论主观还是客观上来看，该算法都具备明显优势。

A Multi-focus Image Fusion Algorithm Based on Multidimensional Adaptive Filtering and Poisson Fusion

In order to solve the problem of the existing multi-focus image fusion algorithms in terms of anti-noise performance, image continuity and time complexity, a multi-focus image fusion algorithm based on multi-dimensional adaptive filtering and Poisson fusion is proposed. First, a gradient matrix is constructed for the source image of uniform scale, and a multi-dimensional adaptive kernel function of scale segment is designed to segment the matrix; then, a judgment condition is introduced to suppress the boundary of the reconstructed image in Poisson fusion. The purified global focus image; finally, the anti-noise algorithm and the fusion algorithm are compared with other algorithm models, and the horizontal comparison is carried out, and the fusion performance and effect are evaluated through 6 image quality evaluation indicators. The results show that the algorithm in this paper not only has good operating efficiency and clear visual perception, but also has more ideal data results in evaluation factors. It can be seen that the algorithm has obvious advantages both subjectively and objectively.