基于两种不同高频融合规则的医学图像融合

Medical image fusion based on two different kinds of fusion rule of high frequency parts

针对多尺度分解医学图像融合时信息损失、分解复杂的问题,提出一种医学图像融合方法。源图像经过图像分解框架分解为高频部分和用于局部拉普拉斯滤波分解的低频部分:第一次高频部分采用感兴趣信息,第二次高频部分采用最大值融合规则,低频部分采用局部能量融合规则,逆局部拉普拉斯滤波重构得到的重构图像和高频融合部分采用加法运算融合得到最终融合结果。实验结果表明,对比几种经典算法,所提方法在客观和主观评价方面显示出其优越性。

To reduce information loss and complex decomposition in multi-scale decomposition medical image fusion,a medical image fusion method was proposed.Source images were decomposed into high frequency parts and low frequency parts used for decomposition of local Laplacian filter by image decomposition frame.High frequency parts were fused by information of interest and second high frequency parts were fused by maximum rule respectively,low frequency parts were fused by local energy maximum rule,the reconstructed image obtained by the inverse local Laplacian filter and the high frequency fusion part were fused by addition operation to get the final fusion result.Compared with several classical algorithms,experimental results demonstrate that its superiority is shown in both objective and subjective evaluation.