利用邻域激励的自适应PCNN进行医学图像融合  

Application of adaptive PCNN based on neighborhood to medical image fusion

作　　者：夏加星[1] 段先华[1] 魏世超[1] 

机构地区：[1]江苏科技大学计算机科学与工程学院,江苏镇江212003

出　　处：《计算机应用研究》2011年第10期3929-3933,共5页Application Research of Computers

基　　金：国家自然科学基金资助项目(60773172);江苏省自然科学基金资助项目(BK2008411)

摘　　要：对于不同模态的医学图像进行融合处理,可为临床提供新的诊断信息,设计了一种邻域空间频域激励的自适应PCNN医学图像融合新方法。首先,使用图像逐像素地改进拉普拉斯能量和(SML)清晰度作为PCNN对应神经元的链接强度;同时利用邻域空间频域(SF)特征信息激励每个神经元;然后,将源图像输入PCNN获得点火映射图构成的点火频数,再判定并选择各参与融合图像中的清晰部分生成融合图像。实验结果表明,该算法具有比经典金字塔方法、小波变换方法和传统的PCNN方法更好的融合性能。The fusion of different medical images can provide new diagnostic information for clinician,this paper introduced a new approach to medical image fusion by using neighborhood spatial frequency inspiring adaptive PCNN.First,this algorithm used the Sum-modified-Laplacian（SML） of each pixel as the linking strength of each neuron,so that the linking strength of each pixel could be chosen adaptively.Meanwhile,neighborhood spatial frequency of the pixels were modeled into a feature information to inspire each neuron.Then,the ignition frequency was obtained via the ignition mapping image generated by the processing of PCNN.The clear objects of each original image were decided by the compare selection operator and then all of them were merged into a new clear image.Experimental results demonstrate that the proposed algorithm has better fusion performance than the classical pyramid,wavelet transform and the conventional PCNN.

关 键 词：医学图像融合 脉冲耦合神经网络 链接强度 改进拉普拉斯能量和 空间频域 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]