ISSN 1004-4140
CN 11-3017/P
刘必成, 易茜, 宗春光, 等. 基于梯度特征聚类分析的大尺寸物体CT图像环状伪影校正方法[J]. CT理论与应用研究(中英文), 2024, 33(6): 779-787. DOI: 10.15953/j.ctta.2024.153.
引用本文: 刘必成, 易茜, 宗春光, 等. 基于梯度特征聚类分析的大尺寸物体CT图像环状伪影校正方法[J]. CT理论与应用研究(中英文), 2024, 33(6): 779-787. DOI: 10.15953/j.ctta.2024.153.
LIU B C, YI X, ZONG C G, et al. Ring-artifact Correction Method for Large-size Object CT Images Based on Gradient Featured Cluster Analysis[J]. CT Theory and Applications, 2024, 33(6): 779-787. DOI: 10.15953/j.ctta.2024.153. (in Chinese).
Citation: LIU B C, YI X, ZONG C G, et al. Ring-artifact Correction Method for Large-size Object CT Images Based on Gradient Featured Cluster Analysis[J]. CT Theory and Applications, 2024, 33(6): 779-787. DOI: 10.15953/j.ctta.2024.153. (in Chinese).

基于梯度特征聚类分析的大尺寸物体CT图像环状伪影校正方法

Ring-artifact Correction Method for Large-size Object CT Images Based on Gradient Featured Cluster Analysis

  • 摘要: 针对CT扫描系统的环形伪影,本文提出一种基于图像梯度特征聚类分析的Hough变换的环状伪影校正方法。利用极坐标图像梯度特征进行聚类分析,对重建图像数据分类初筛,将重组数据采用Hough变换直线检测并映射找出原图像中的伪影数据,进而插值消除极坐标图像的直线伪影,达到重建图像环形伪影消除的目的。模拟和实测图像的测试结果表明,采用图像梯度特征提取后,检测出有效伪影直线的数量大幅提升,重建图像中的环状伪影、弧形伪影及中心区的团状伪影都得到有效去除,且大大减少传统Hough变换法误检测直线的现象。

     

    Abstract: In this paper, a Hough transform method based on image gradient featured extraction is proposed to eliminate the ring artifacts of CT scanning system. Firstly, we use the gradient featured cluster analysis of the polar coordinate image to preliminary select the reconstructed image data. Then the Hough transform is adopted to detect the lines of the recombined data. When the artifacts in the original image are mapped out, the linear artifacts in the polar coordinate image are eliminated by interpolation. Then we can purposefully eliminate the ring artifacts in the reconstructed image. Simulation and experimental results show that with the image gradient featured cluster analysis, the effective number of detected artifact lines is greatly increased, the ring artifacts, arc artifacts and cluster artifacts in the central area of the reconstructed images are effectively removed, and the erroneous detection of lines by traditional Hough transform method is greatly reduced.

     

/

返回文章
返回