ISSN 1004-4140
CN 11-3017/P
吕振涛, 桂志国, 刘祎, 张鹏程. 基于加权平均的工业CT旋转中心校正方法[J]. CT理论与应用研究, 2021, 30(5): 539-546. DOI: 10.15953/j.1004-4140.2021.30.05.01
引用本文: 吕振涛, 桂志国, 刘祎, 张鹏程. 基于加权平均的工业CT旋转中心校正方法[J]. CT理论与应用研究, 2021, 30(5): 539-546. DOI: 10.15953/j.1004-4140.2021.30.05.01
LV Zhentao, GUI Zhiguo, LIU Yi, ZHANG Pengcheng. Industrial CT Rotation Center Calibration Method Based on Weighted Average[J]. CT Theory and Applications, 2021, 30(5): 539-546. DOI: 10.15953/j.1004-4140.2021.30.05.01
Citation: LV Zhentao, GUI Zhiguo, LIU Yi, ZHANG Pengcheng. Industrial CT Rotation Center Calibration Method Based on Weighted Average[J]. CT Theory and Applications, 2021, 30(5): 539-546. DOI: 10.15953/j.1004-4140.2021.30.05.01

基于加权平均的工业CT旋转中心校正方法

Industrial CT Rotation Center Calibration Method Based on Weighted Average

  • 摘要: 旋转中心的精确定位是重建高质量CT图像的一个重要条件。本文提出一种基于加权平均的旋转中心校正方法,该方法首先将投影正弦图以180°为界限分成两组投影数据,计算出这两组投影数据的互相关系数及其对应的探测器探元坐标,然后使用最大与次最大的互相关系数对探测器探元坐标值进行加权平均得到实际旋转中心坐标。本文所提算法完全基于原始投影数据,无需确定CT测量的几何参数,不用提前对投影数据进行插值运算就可以将旋转中心精确到亚像素级,精度更高,实现简单。

     

    Abstract: The precise positioning of the rotation center is an important condition for reconstructing high-quality CT images. In this paper we propose a center rotation calibration method based on weighted average. First we divide the projection sinogram into two sets of projection data by limit of 180°, thus we can figure out the calibration coefficient of the two sets of projection data and their corresponding detector probes. Then we use the maximum and the sub-maximal cross-correlation coefficients to obtain the actual rotation center coordinates by weighting average the coordinate values of the detector elements. The algorithm proposed in this paper is completely based on the original projection data. There is no need to determine the geometric parameters of the CT measurement, and we can acquire sub-pixel level accuracy of the rotation center without interpolating the projection data in advance. The accuracy is higher while the implementation is quite simple.

     

/

返回文章
返回