一种新的基于多尺度几何分析的图像融合方法

基于多尺度几何分析方法——非下采样轮廓波(Con tourlet)变换(NSCT)和Beamlet变 换,提出一种全新的医学图像融合方法。在进行NSCT分解后,在高频成分首先使用Beamlet 变换进行边缘检测, 然后根据聚类分割边缘密度的差值确定其系数的融合规则；对于低频成分,采用局部区域标 准方差系数的融合规则；经过 一致性校正后,通过对融合后的高频与低频子带系数进行逆NSCT得到重构图像。数值实验表 明,与传统的融合方法相 比较,本文方法能够有效减少噪声对融合图像的干扰,增强了融合的线性细节表达能力,提 高了信息量。

A novel medical image fusion method based on the multi-scale geometric analysis tool

This study proposes a novel medical im age fusion method based on the multi-scale geometric analysis tool-non-subsam pled contourlet transform (NSCT) and Beamlet transform.The NSCT is applied to im age processing field because of its directional,anisotropic and translational in variance properties.Beamlet transform has the advantage of perfect line feature detection ability.At first,this algorithm de composes the images by the NSCT.In high frequency region,it uses the Beamlet transform to make edge detection.Then it uses the edg e density difference value of the clustering segmentation to get the coefficient fusion rules.In low frequency region,this algorithm uses the standard variance coefficient of partial region to get the fusion rules.After that,it performs consistency correc tion for the fused coefficients.Finally,it gets the reconstructed image by the inverse NSCT transform on the fused high frequency and low fr equency subband coefficients.The experiment results show that compared with traditional fusion methods,this algorithm can e ffectively diminish the fusion image ambiguity caused by the noise.It enhances the linear details presentation ability and inc reases the information amount.