摘要
为了使PET/CT医学图像融合达到更好的效果,讨论了小波变换和区域的最优融合方法。首先将PET图像和CT图像通过二维离散小波变换,得到低频子带和高频子带;其次采用邻域能量、邻域方差、邻域梯度3种算法对高频子带进行规范化,融合规则均选用绝对值取大的方法;然后选用取平均值的方法对低频子带进行融合,将融合后的PET/CT图像通过二维离散小波逆变换,获得最终的融合图像;最后对3种融合算法进行比较,得出最优算法。实验结果表明,基于邻域梯度的融合方法在客观评价指标和人眼视觉效果上均优于其它方法,且鲁棒性较强。
In order to obtain a better fusion effect of PET/CT medical image, the fusion method based on wavelet transform is discussed in this paper. Firstly, the original image are decomposed to low and high frequency sub-images using two-dimensional discrete wavelet transform; Secondly, there Kinds of algorithms are used for standardization in the high frequency components of the image, such as neighborhood energy, neighborhood variance and neighborhoodgradient,which a method based on maximum value was utilized to fuse the high-frequency componentsrespectively; Thirdly, a fusion rule for computing mean value is adopted in low frequency sub-band. Then, the final fusion image is acquired through two dimensional discrete wavelet inverse transform; Finally, the optimal algorithm is obtained via the comparison of three kinds of fusion algorithm based region. The experimental results show that the fusion algorithm based on neighborhood gradient is superior to other algorithms in terms of human visual effect and objective evaluation, and it has strong robustness
作者
陈琛
霍兵强
李海军
陆惠玲
杨旭东
CHEN Chen HUO Bing-qiang LI Hai-jun LU Hui-ling YANG Xu dong(School of Science, Ningxia Medical University, Yinchuan 750004, China School of Computer Science and Technology, Harbin Engineering University, Harbin 150001, China The Center of Public Management, Ningxia Medical University, Yinchuan 750004, China)
出处
《软件导刊》
2017年第10期202-204,208,共4页
Software Guide
基金
宁夏高等学校科研项目(NGY2016084)
宁夏自治区大学生创新创业计划项目(NXCX2015204)