Abstract: A calibration method is proposed, based on the analysis of defects causes and characteristics in projective images produced by cone beam CT with Flat-Panel Detector（FPD）, to correct the dark field, response nonuniformity in one step, and to fit the light field and remove the bad pixels. Acquire sequence images of dark-field and air projection in different amounts of exposure in case of the same voltage level in the same cone-beam CT system, and then analyze its pixel value variation, fitting and contrast, to get the linear parameter matrix of the system, the light field image and the bad pixel map. Correct the air projection image, and treat it as an incomplete reference image to calculate the x-ray absorption coefficient for cone-beam CT reconstruction. The results show that the method effectively eliminates the noise in cone-beam CT projection images and inhibits the number of ring artifacts in object reconstruction slice. It can improve the quality of cone-beam CT system reconstruction image.

Key words: cone-beam CT, Flat-Panel Detector（FPD）, projective image correction, image intensifier

摘要： 在分析基于平板探测器的锥束CT投影图像缺陷的产生原因和特性的基础上，提出校正的方法，分别对投影图像进行暗场、像元响应不一致的一步校正和光场不均匀校正、像素坏点去除。通过在同一锥束CT系统中相同电压、电流水平情况下，采集不同积分时间下的空气投影图和相同条件下的暗场图，并对其像素值变化进行分析、拟合和对比，得到系统的线性参数矩阵、光场分布图和坏点位置图。对空气投影图像进行校正，并将其作为不完全参照，计算射线吸收系数，用于锥束CT重建。实验结果表明，该方法能有效去除锥束CT投影图像的噪声，抑制物体重建切片的大量环状伪影的产生，提高锥束CT系统重建图像的质量。

关键词: 锥束CT, 平板探测器, 投影图像校正, 图像增强