固体火箭发动机CT图像条状伪影校正

为了校正窄角扇束工业CT扫描重建的固体火箭发动机图像中的条状伪影，分析了条状伪影的产生原因.针对条状伪影是由重排和数据对齐后固体火箭发动机投影数据的非光滑连续性引起的，提出了条状伪影校正的两种算法.两种算法均以中间平移步数的投影数据为基准进行投影数据的重排，一种将重排和数据对齐过程分开处理；另一种将重排和数据对齐过程合并处理.为了验证两种条状伪影校正算法的有效性，采用窄角扇束工业CT对某型固体火箭发动机进行3种不同扫描步长的扫描和重建.结果表明重排和数据对齐过程分开处理的校正算法只有在小步长扫描的情况下重建效果较好，条状伪影较少；而重排和数据对齐过程合并处理的校正算法在3种不同的扫描步长下图像重建质量均较好，条状伪影较少. 

Streak artifacts correction of solid rocket motor's CT image

To correct streak artifacts in the computed tomography (CT) image of solid rocket motor (SRM), scanned by narrow fan-beam industrial CT (ICT), the reason for the streak artifacts was analyzed. Considering the streak artifacts were caused by the inconsistency of projection data of SRM after rebinning and data shake-up processes, two algorithms for the correction of streak artifacts were proposed. The two algorithms used the projection data in the half of the translation steps as the reference for the rebinning of projection data. For one algorithm, rebinning process and data shake-up process were considered as two processes. For the other, rebinning and data shake-up processes were considered as one process. To prove the effectiveness of the two algorithms, one SRM was scanned by narrow fan-beam ICT with three different scanning step sizes and SRM CT images were reconstructed by the two algorithms. Results show that the algorithm using rebinning and data shake-up as two processes performs well only if the scanning step size is small, and there are few streak artifacts in this situation. The other algorithm using rebinning and data shake-up as one process performs well and generates few streak artifacts in the situations of three different step sizes.