摘要
工业计算机断层成像(CT)扫描大尺寸和高密度工件时,会出现穿不透、扫描角度有限导致的投影数据不完备、重建伪影严重等问题。基于此,提出一种将工件的CAD设计模型作为先验知识,来实现该类对象的有限角CT重建优质图像的方法。由工件CT扫描的断层位置计算CAD模型对应的分层位置,并得到模型的像素截面;根据CT系统X射线能量和模型中各部分材质,确定和生成一幅衰减系数图像,并将其配准到一幅零灰度图像中,得到先验图像;最后对扫描投影数据进行SART+TVM+PRIOR算法重建,得到重建图像。仿真实验和实际工件扫描实验的结果显示,加入先验图像后重建的图像质量要远远优于未加入先验图像的重建结果,边缘结构更加清晰,并极大地减少了伪影。
Large-size, high-density workpieces used in industrial computed tomography(CT) scanning can cause problems such as incomplete projection data and terrible artifacts in reconstructed images owing to a lack of penetration and limited scanning angles. To address such limitations, this study uses a CAD model of the workpieces as prior knowledge and realizes superior reconstruction of the limited CT angle occurring in these types of objects. The layered position corresponding to the CAD model is calculated by the tomography workpiece position of the CT scan, and the pixel section of the model is obtained. Then, an attenuation coefficient image is obtained and generated according to the X-ray energy of the CT system and material of each part of the model that is registered into a zero gray image to obtain a prior image. Finally, the simultaneous algebraic reconstruction technique+total variation minimization+prior image(SART+TVM+PRIOR) algorithm is used to iterate the scanned projection data to obtain the reconstructed image. The experimental results of the simulation and actual workpiece scanning indicate that the quality of reconstructed image with the prior image is significantly improved over that without the prior image. This improvement is reflected mainly by a substantial reduction in artifacts in the clearer edge structure.
作者
余浩松
邹永宁
张智斌
姚功杰
周日峰
Yu Haosong;Zou Yongning;Zhang Zhibin;Yao Gongjie;Zhou Rifeng(Engineering Research Center of Indlustrial Computed Tomography Nondestructive Testing,Ministry of Edrcation,Chongqing University,Chongqing 400044,China;College of Optoelectronic Engineering,Chongqing University,Chongqing 400044,China)
出处
《光学学报》
EI
CAS
CSCD
北大核心
2021年第6期101-111,共11页
Acta Optica Sinica
基金
国家自然科学基金(11827809)。