基于探测器响应机理的碳/碳构件CT图像环状伪影的校正方法

A Correction Method of Ring Artifact in CT Image of C/C Component Based on Detector Response Mechanism

在碳/碳复合材料的计算机断层扫描(CT)中,由于探测器单元的响应非线性及不一致性导致重建图像出现严重的环状伪影,干扰图像中缺陷的检出,影响检测系统对被检构件的质量评价。结合碳/碳复合材料组分单一且密度均匀的特点,提出了一种基于探测器响应机理的CT图像环状伪影的校正方法,利用采集的劣化投影数据做预重建,对重建结果做阈值分割以获得被检测物体的三维模型。结合已知的材料和密度信息,对被检物体进行重投影,得到投影数据的理论值与实测值之间的映射关系,用于探测器响应校正以优化检测图像。与传统低通滤波的环状伪影校正方法相比,该方法考虑了环状伪影的物理成因并充分利用了检测对象的先验信息。研究结果表明,该校正方法在保留图像细节和纹理的同时,能够有效减少环状伪影,提升图像质量,消除伪影对于图形中缺陷识别的干扰,为提升检测系统对碳/碳复合材料构件缺陷的检出能力提供理论依据。

In computed tomography(CT)imaging of C/C composite materials,one of the main challenges is the presence of significant ring artifacts in the reconstructed image.These artifacts are primarily caused by the non-linear and non-uniform response of the detector,which hinders the accurate detection and evaluation of defects in the inspected components.To overcome this issue,a novel physical correction method based on the underlying detector response mechanism was proposed in this work.The key idea behind the proposed method is to take advantage of the homogeneous composition and uniform density of C/C composite materials.Firstly,a pre-reconstruction process was conducted to locate the shape and position of C/C components utilizing the degraded projection data.Moreover,a threshold segmentation technique was employed to extract a detailed and accurate three-dimensional model of the inspected C/C components.By performing forward projection on the segmented three-dimensional model of the inspected C/C components,theoretical projection data of homogeneous components could be obtained.The next step involved combining the available information about the material and density to carry out a re-projection procedure.This analysis enables the establishment of a robust mapping relationship between the theoretical and measured values of the projection data,which is then employed to correct for the detector response by a polynomial fitting procedure.Finally,the corrected projection data was used to reconstructed the optimized CT images of which ring artifacts would be significantly suppressed.Comparing with conventional low-pass filter based methods for addressing ring artifacts,this approach takes into account the physical causes of these artifacts while making full use of the prior knowledge regarding the inspected object.Not only does it effectively reduce the presence of ring artifacts,but it also significantly retains the texture and details of the original CT images.By eliminating the interference caused by these artifacts,the method enhances the detection system’s ability to identify and evaluate defects accurately.Consequently,this technique provides a valuable theoretical foundation for improving the defect detection capabilities of CT inspection systems for C/C components.It successfully addresses the challenges posed by ring artifacts by considering their physical origins and leveraging the prior knowledge about the inspected object.The results of the study demonstrate its effectiveness in reducing ring artifacts,enhancing image quality,and eliminating interference during defect identification.Moving forward,this method holds great potential for further improving the performance and reliability of CT inspection systems for C/C components.