中文摘要：

针对各类成像传感器生成图像具有不同的特征信息和医学图像特定应用需求,提出了一种基于提升小波变换和PCNN的多模态医学图像融合算法。首先对已预处理源图像进行提升小波分解获得高、低频子带;其次根据低频部分区域方差判定图像区域相关性,并以区域能量获取系数权重;然后对高频部分采用改良空间频率刺激PCNN网络,并以点火区域强度作为系数判定标准;最后对融合后所得子带通过提升小波逆变换重构获得融合图像。实验结果表明,该算法在信息熵、标准差和边缘传递因子3个指标上有较大提升,较好地保留了边缘细节信息,融合后的图像信息比传统算法更丰富。

英文摘要：

Since the images generated by various imaging sensors have different feature information and medical images have specific application demand,this paper proposes a multi-modal medical image fusion algorithm based on lifting wavelet transform and PCNN.Firstly,the source image preprocessed was decomposed by lifting wavelet transform into high and low frequency sub-bands.Secondly,image region correlation was j udged according to regional variance of low frequency part,and the coefficient weight was obtained based on the regional energy.Thirdly,for high frequency part,improved spatial frequency was applied to stimulate PCNN network,and ignition regional intensity was used as coefficient j udgment criterion.Finally,fused image was gained through lifting wavelet inverse transformation for fused sub-bands.The results show that the algorithm improves greatly in terms of 3 indicators including information entropy,standard deviation and edge transfer factor,and well reserves edge details.The fused image information is richer than that of traditional algorithm.