基于非线性估计的CBCT几何标定算法

CBCT geometric calibration algorithm based on nonlinear estimation

针对传统的几何标定算法由标定模体位姿误差导致的CBCT成像系统参数标定精度不高的问题,提出一种基于解析几何和非线性优化的CBCT几何标定算法。首先,设计一种插帧追踪的方法,提取特征点的坐标,使得标定模体与投影图像中的特征点实现一一对应。然后,利用投影图片与设计的标定模体之间的空间几何关系,计算相关参数。最后,对检测的特征点进行非线性估计,进一步提高几何参数的计算精度。仿真数据实验表明,本文算法能够较为精确地计算出CBCT成像系统的几何参数,并在标定模体出现位姿偏差时仍然保持较高的计算精度。通过非线性估计模型,本文所提出的算法突破了现有算法标定模体位于等中心点的限制,提高了CBCT成像系统中参数标定的精度。

Due to the posture error of calibration phantom,traditional geometric calibration algorithm has a poor accuracy in the parameter calibration of cone beam computed tomography(CBCT)imaging system.In order to solve the abovementioned problem,a CBCT geometric calibration algorithm based on analytic geometry and nonlinear optimization is proposed.A tracking method based on frame interpolation is firstly designed to extract the coordinates of feature points,thereby achieving the correspondence between the calibration phantom and the feature points in the projected image.Then the relevant parameters are calculated based on the spatial relationship between the projected image and the designed calibration phantom.Finally,the nonlinear estimation of the detected feature points is carried out to further improve the accuracy of geometric parameter calculation.The simulation results verified that the proposed algorithm can accurately calculate the geometric parameters of CBCT imaging system,and maintain a high calculation accuracy even when the posture error is occurred during phantom calibration.Based on the proposed nonlinear estimation model,the proposed method overcomes the limitation of the existing algorithm in which calibration phantom needs to be located at the isocenter and improves the accuracy of parameter calibration in CBCT imaging system.