基于预分割轮廓的CT图像亚体素表面检测方法  

Subvoxel-accuracy surface detection method based on contour pre-segmentation for computed tomography images

作　　者：查方龙[1] 张定华[1] 黄魁东[1] 张亮[1] 李明君[1] 

机构地区：[1]西北工业大学现代设计与集成制造技术教育部重点实验室,西安710072

出　　处：《仪器仪表学报》2012年第6期1308-1314,共7页Chinese Journal of Scientific Instrument

基　　金：国家科技重大专项(2012ZX04007-021);国家自然科学基金(51105315);西北工业大学种子基金(Z2011080)资助项目

摘　　要：针对基于CT(computed tomography)图像检测分析中的点云提取精度与完整性问题,提出一种基于预分割轮廓的高精度、高完整性的亚体素表面检测方法。首先采用Otsu分割算法提取CT图像的体素级轮廓点集,并以此作为粗定位轮廓自适应地生成用于亚体素表面检测的完备感兴趣区域(region of interest,ROI);然后提出一种基于梯度非极大值抑制的表面体素判定方法,避免了梯度阈值选择难题;最后基于3D Facet模型定位亚体素级表面点位置。实验结果表明,该方法能有效改善传统亚体素检测方法的轮廓丢失、伪边严重等问题,轮廓定位误差小于0.2个体素,同时能够取得3倍以上的计算加速比。Aiming at the precision and integrity problems of point cloud extraction in the detecting applications based on computed tomography（CT） images, this paper proposes a pre-segmentation based subvoxel-accuracy surface de- tection method with fine integrity and high precision. Firstly, the Otsu segment algorithm is adopted to obtain the ini- tial sets of voxel-accuracy contour points for the CT image. With these sets as the coarse positioning contour, the complete region of interest（ROI） for subvoxel-accuracy surface detection is adaptively generated. Then, a surface voxel judging criterion is put forward based on non-maximum gradient suppression strategy, which avoids the gradient threshold selection dilemma. Finally, the positions of the subvoxel-accuracy surface points are determined based on 3D Facet model. Experiment results indicate that our method has a significant promotion in overcoming the contour loss and severe pseudo edges. The total positioning precision could be less than 0.2 voxels, and it could also obtain a computational speedup ratio above 3 compared with conventional methods.

关 键 词：CT图像 表面检测 自适应感兴趣区域 3D FACET模型 亚体素精度 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]