基于多聚类中心和PCNN的医学图像融合算法

摘要

提出一种基于K-means Clustering和脉冲耦合神经网络（PCNN）的图像融合的方法，首先，以多特征信息为聚类方式利用K-means Clustering分割提取源图像的对应特征点，通过归类合并建立多模医学图像的特征点集合，根据特征点分布将图像划分为纹理区域和非纹理区域，纹理区域对应系数输入PCNN得到点火映射图，根据点火次数选择融合系数，非纹理区域的系数通过双通道PCNN进行融合。实验结果表明，该算法能够精确划分图像纹理区域，进而利用PCNN和双通道PCNN在图像不同区域系数选择各自的优势，融合图像纹理清晰，质量改善。%A medical image fusion method is proposed based on K-means Clustering and pulse coupled neural network (PCNN). Firstly,K-means Clustering method is used to segment the source image based on multi-feature information and the corresponding feature points are extracted,the multimodal medical image feature points are classified and mer-ged to build feature points set. According to the distribution of feature points,the images are divided into texture re-gions and non-texture regions. The coefficients of texture region are input into PCNN and the ignition map is ob-tained;According to the ignition frequency,fusion coefficients are selected. The fusion coefficients of non-texture re-gion are based on dual-channel PCNN. Experimental results show that the algorithm can accurately segment image texture region,and thus take advantage of dual-channel PCNN and PCNN in coefficient selection in different regions of the image. The proposed method gets a better image fusion result.