固体火箭发动机窄角扇束CT投影直接重建算法

摘要

用于固体火箭发动机检测的窄角扇束 CT 采用重排算法进行图像重建，而重排算法用到的插值运算会造成投影数据的不准确，使得重建的 CT 图像质量下降。为了解决重排算法存在的问题，提出了一种新的直接卷积反投影算法。该算法避免了重排过程，直接利用不等间距平行束投影数据进行卷积反投影重建。为了验证直接卷积反投影算法的有效性，进行了数值仿真和实验验证，并将直接卷积反投影算法与重排算法进行对比。仿真和实验结果表明，直接卷积反投影算法比重排算法的图像重建时间短，直接卷积反投影算法较重排算法的重建图像质量高，这对固体火箭发动机缺陷的精确测量具有重要意义。%Rebinning algorithm is applied to reconstruct image by narrow fan⁃beam CT for the inspection of solid rocket motor (SRM).However,the interpolation process used in rebinning algorithm causes inaccuracy of projection data,which reduces the qual⁃ity of reconstructed CT image.To solve the problem of rebinning algorithm,a new direct convolution backprojection (DCBP) algo⁃rithm is proposed.The DCBP algorithm avoids rebinning process and reconstructs CT image with unequispaced parallel⁃beam projec⁃tion data directly.To test the effectiveness of the DCBP algorithm and compare it with rebinning algorithm,numerical simulations and experiment were conducted. Results show that the computation time of the DCBP algorithm is less than that of rebinning algorithm, while the quality of CT image of the DCBP algorithm is better than that of rebinning algorithm.This is of great significance for the precise measurement of defects of SRM.