Abstract: In order to better meet the modern medicine clinical diagnosis and treatment needs, improve the medical image fusion quality, this paper proposes a new medical image fusion algorithm based on lifting wavelet transform, combined with the Pulse Coupled Neural Networks（PCNN） and pixel point of nonlinear filtering gravitation. A fusion rule based on the area average gray is adopted in low-frequency coefficients. Adaptive PCNN fusion rule is adopted in high-frequency coefficients. It makes the pixel’s nonlinear filtering gravitation as the linking strength of simplified PCNN model. So PCNN can adaptively control the size of linking strength, and according to the ignition matrix to determine the fusion of high-frequency coefficients. Experiments show that the fused image obtains more edge detail information than other methods and every index has improved. It has a better fusion performance.

Key words: medical image fusion, lifting wavelet transform, Pulse Coupled Neural Networks（PCNN）, nonlinear filtering gravitation, linking strength

摘要： 为了更好地满足现代医学临床诊断和治疗的需要，提高医学图像的融合质量，提出在提升小波变换的基础上，结合脉冲耦合神经网络（PCNN）和像素点的非线性滤波万有引力的医学图像融合方法。低频子系数采用基于区域灰度均值的融合规则；高频子系数采用自适应PCNN的融合规则，将像素的非线性滤波万有引力作为简化的PCNN模型中的链接强度，使PCNN能够自适应调节链接强度的大小，并根据点火矩阵确定高频子系数。实验结果表明，该方法得到的融合图像比其他融合方法保留了更多的边缘细节信息，各项评价指标均有所提高，有更好的融合性能。

关键词: 医学图像融合, 提升小波变换, 脉冲耦合神经网络, 非线性滤波万有引力, 链接强度