为了提升医学图像融合质量, 采用了一种基于2维经验模态分解(BEMD)特征分类和复合型脉冲耦合神经网络的医学图像融合算法。首先将多模医学图像经过BEMD分解成2维内蕴模函数(BIMF)和残差项, 然后分别将BIMF层和残差项值输入脉冲耦合神经网络(PCNN)中, 得到各自的点火映射图, 再将相同点火次数的像素提取归类, 点火次数大的对应图像纹理, 归为纹理类, 其余归为背景类; 统计各个纹理类集合中的像素极值确定灰度分布范围, 最后将两幅图像中纹理类像素集合处于灰度分布范围的像素通过PCNN进行融合, 其它像素通过双通道PCNN进行融合。结果表明,该算法解决了PCNN对偏暗图像的处理效果不理想的问题, 与传统融合算法相比, 性能具有优势,且能够较大幅度提高融合图像的质量。

In order to improve the quality of medical fusion images, a novel medical image fusion algorithm based on bidimensional empirical mode decomposition (BEMD) feature classification and multi-pulse coupled neural network was proposed. Firstly, the multimodal medical images were decomposed into two-dimensional intrinsic mode functions (BIMF) and the residuals by means of BEMD, and then the BIMF layer and the residuals coefficients were put into pulse coupled neural network (PCNN) to get their firing maps. The pixels with the same firing times were extracted and classified. The pixels with larger firing times were classified as texture and the rest were classified as the background. The extreme values of the texture collection were counted to determine the grayscale pixel distribution. Finally the pixels representing the texture were input into the PCNN and the other pixels were put into the dual-channel PCNN to get fusion coefficients. The experimental results show that the proposed algorithm has solved the problem of PCNN with superior performance comparing to the traditional fusion algorithms, which can improve the quality of the fused image.