摘要
提出了一种基于蒙特卡罗积分的数字影像重建方法,首先根据概率密度函数采样形成点云集合,并将空间采样转换为强度采样,采样速度获得了较大提升;然后利用空域滤波器平滑点云集,避免投影出现孔洞;最后根据不同的视角对样本投影,统计出像平面上每个区域内的投影数目.为降低图像估计方差,提出结合Russian roulette技术的混合采样方法和Metropolis采样方法.相比较于复杂度为O(N3)的确定性投影算法,所提出的算法复杂度降为O(N2).实验结果表明,该方法能以较快的帧速给出大数据集的数字影像重建,且无需多次调整转换函数即可生成类X射线影像.
A digitally reconstructed radiography（DRR） method based on Monte Carlo integral is presented.The point cloud set is formed firstly according to the probability density function by sampling the intensity instead of the position and the speed of sampling has been improved much more.Then the space filter is applied to smooth the point cloud to avoid the projecting holes.Finally the number of the projecting points is counted from the different views on each area of the image plane.Russian roulette and Metropolis techniques are utilized to decrease the variance of generated image.Compared with the deterministic methods, the algorithm complexity is decreased from O（N^3） to O（N^2） for the present method.The experiments demonstrate that the proposed method can obtain the DRR with a higher frame speed for the large data set and without more adjustment for the transfer function.
出处
《东南大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2008年第6期992-995,共4页
Journal of Southeast University：Natural Science Edition
基金
国家重点基础研究发展计划(973计划)资助项目(2003CB716102).
关键词
蒙特卡罗积分
数字影像重建
重要性采样
Monte Carlo integral
digitally reconstructed radiography
importance sample
作者简介
沈傲东（1978-），男，博士生；
罗立民（联系人），男，博士，教授，博士生导师，luo．list@seu．edu．cn．
