多特征结合引导滤波的多聚焦图像融合

Multi-Focus Image Fusion based on Multiple Features Combined with Guided Filtering

提出一种多特征结合引导滤波的图像融合算法,使用多个特征,计算各特征的置信图,将各特征的优势相结合,提升整体融合的质量。具体步骤如下:首先计算源图像的多类特征,各特征经过引导滤波后计算初始决策图和置信图,然后通过形态学方法去除小连通区域后,再次使用引导滤波得到最终决策图,最后使用置信图对最终决策图加权,与源图像相乘再相加得到最终的融合图像。实验结果表明,该算法对残差图的主观分析和客观融合质量评价指标的定量分析,较当前传统多聚焦图像融合算法均有提升。

An image fusion algorithm with multiple features combined with guided filtering is proposed,which uses multiple features,calculates the confidence map of each feature,combines the advantages of each feature,and improves the quality of the overall fusion.The specific steps of the algorithm are as follows:first,calculate multiple-type features of the source image,and the initial decision map and confidence map after each feature is guided and filtered;then after removing the small connected regions by morphological methods,the guided filtering is used again to obtain the final decision graph;finally,the confidence map is used to weight the final decision map,multiplied by the source image and then added to obtain the final fused image.The experimental results indicate that the subjective analysis of the residual image and the quantitative analysis of the objective fusion quality evaluation index are both improved as compared with the current traditional multi-focus image fusion algorithms.