摘要
由于成像机理不同和人体组织结构的高度复杂性,单模态的医学图像不能提供医生所需要的足够信息,因此对于多模态医学图像的配准和融合有着十分重要的意义。基于此以互信息作为相似性测度,改进单纯形法为最优化算法对CT和MRI图像进行配准。采用了小波变换对图像进行融合。高频部分,取两幅图像小波系数矩阵对应元素的最大绝对值构造小波系数矩阵;低频部分,采用了基于领域像素相关和基于区域方差相结合的融合策略。结果表明,融合图像清晰,细节丰富,相对位置准确,反映了原始图像中更为全面的、互为补充的多重信息。
Because of the different imaging mechanism and high complexity of body tissues and structures, the single-modality medical image cannot provide enough information for clinical doctors. Different modality medical images provide non-overlay comple- mentary information. For the two image registrations of MR/and CT, the modified simplex algorithm is employed to implement optimal algorithm with the using of mutual information as similarity measure. After registration the medical images are fusion by wavelet trans- forms. For high frequency fusion, the new coefficients are selected by those coefficients with maximum absolute values in two original images. For low frequency fusion, it is used to combine with the strategy on domain pixel correlation and regional variance. The result shows that the fusion image is clear, details are abundant, relatively position is accurate. It reflects that multi-information is the more overall and supplementing each other.
出处
《山西电子技术》
2013年第2期89-91,共3页
Shanxi Electronic Technology
基金
山西省青年科技研究基金(2012021011-1)
关键词
图像配准和融合
互信息
改进单纯形
小波变换
image registration and fusion
mutual information
modified simplex algorithm
wavelet transform
