SML与侧窗滤波的Framelet域医学图像融合方法

Medical Image Fusion Method Using SML and SideWindow Filter in Framelet Transform

针对医学融合图像中病灶信息不清晰的问题,提出了一种框架变换(framelet transform,FT)域的医学图像融合方法。针对待融合医学源图像进行FT变换,分别获得一幅低频子带图像和一系列高频子带图像;构建改进型拉普拉斯能量和模型,实现低频子带图像中主体信息的融合,并得到对应的低频子带融合图像;构建改进型侧窗滤波模型,实现高频子带图像中的细节和纹理信息的有效融合,并获得高频子带融合图像;针对低频子带融合图像和高频子带融合图像进行FT逆变换,从而获得最终融合结果图像。仿真实验结果表明,与近年的代表性方法相比,基于该方法生成的融合图像无论在主观视觉效果还是客观评价指标方面均具有显著的优势。

A fusion method on medical images in framelet transform(FT)domain is proposed to address the issue of unclear lesion information in medical fusion images.Firstly,FT is performed on the medical source image to be fused to obtain a low-frequency sub-band image and a series of high-frequency sub-band images,respectively.Secondly,an improved sum of modified Laplacian model is constructed to achieve the fusion of the main information in low-frequency sub-band images,and the corresponding low-frequency fused sub-band image can be obtained.Thirdly,an improved side window filter(SWF)model is devised to effectively fuse the details and texture information in high-frequency sub-band images,and a series of high-frequency fused sub-band images can be generated accordingly.Finally,the inverse FT is performed on the above low-frequency and high-frequency sub-band fused images to achieve the final fused image.The simulation experimental results show that,compared with the recently representative methods,the fused image based on the proposed method has significant superiorities in terms of both subjective visual effects and objective evaluation indicators.