摘要
针对医学诊断实时传输需求,提出一种基于压缩感知的医学图像融合研究算法.该算法采用基于压缩感知的融合方法,对非下采样轮廓波变换(NSCT)分解得到的高频子带用随机高斯矩阵分别进行测量,利用正交匹配追踪法重建各高频子带系数,对得到的高频部分和低频部分进行逆变换得到融合图像.通过对CT与PET肝脏医学图像的仿真实验,该算法可以增加多模态医学图像互补信息,并能较好地提高医学图像融合的清晰度.比较融合评价指标,证实本算法在保证融合质量的同时有效提高了运算效率,有利于满足医学诊断实时传输的应用需求.
According to the medical diagnosis and real-time transmission demand, a kind of research on medical image fusion algorithm based on compressed sensing is put forward. The algorithm uses the fusion method based on compressed sensing and uses the random Gauss matrix to measure the high frequency sub bands respectively which are obtained by non-subsampled contourlet transform. Then it uses the orthogonal matching pursuit method to reconstruct the high frequency subband coefficients, and finally uses the inverse transform of NSCT to get the fusion image of the high frequency part and the low frequency part. Through the simulation experiment on CT and PET in liver medical images, the algorithm can increase the complementary information of multimodality medical image, and improve the clarity of the medical image fusion. The fusion evaluation indexes show that this algorithm can improve the calculation efficiency effectively while ensuring the fusion quality, and is good to meet the needs of real-time transmission applications in medical diagnosis.
出处
《广西科技大学学报》
CAS
2015年第3期31-35,共5页
Journal of Guangxi University of Science and Technology
基金
国家自然科学基金项目(61302178)资助
关键词
医学图像
非下采样轮廓波变换
压缩感知
图像融合
medical image
non-subsampled contourlet transform
compressive sensing
image fusion
