摘要
在采用面探测器阵列进行三维直接重建CT图像的时候,经常出现大量的线状伪像,对于图像质量和真正缺陷的识别影响很大。这些伪像大多是由投影数据中的随机疵点引起的。文章在雷当域用疵点邻域的均值代替疵点数值的方法来处理CT图像的线状伪像。试验结果证明这种方法简单有效,几乎能去除所有的线状伪像,提高了图像质量,对重建的时间几乎没有影响。
Lots of streak-type artifacts exist in 3D CT (Computed Tomography) images after reconstruction, which do much harm to image quality and true flaw. After careful study, we found out these artifacts is caused by random defects in projections. This paper puts forward a method called relative threshold in Radon domain, which eliminates these artifacts by eliminating random defects in Radon domain. Experiment results proved this method is quite effective. It can eliminate almost all streak-type artifacts, improve image quality, and increase little to reconstruction burden (time cost).
出处
《核电子学与探测技术》
CAS
CSCD
北大核心
2009年第1期101-104,共4页
Nuclear Electronics & Detection Technology
基金
国家自然科学基金60602051资助
关键词
CT
线状伪像
疵点
面阵探测器
CT
Streak-type artifacts
Defect
Area detector