工业CT图像亚体素表面检测算法研究  

Research on Subvoxel Surface Detection Algorithms of ICT Images

作　　者：王凯[1] 张定华[1] 赵歆波[1] 黄鹤龄[2] 刘晶[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 [2]西北工业大学计算机学院,西安710072

出　　处：《CT理论与应用研究（中英文）》2005年第3期40-45,共6页Computerized Tomography Theory and Applications

基　　金：自然科学基金(项目编号50375126);航空科学基金(项目编号04I53069)资助

摘　　要：针对ICT图像序列,研究了基于Facet模型和基于矩的亚体素表面检测算法,并通过引入基于Otsu的阈值分割预处理环节,大大减少了待处理体素的数目,在很大程度上提高了原始算法的处理速度。最后在对航空发动机叶片仿真数据的实验中,对比了算法处理效果,结果表明两算法检测精度均可达1/5个像素以内,预处理环节的引入可将原始算法速度提高约4倍。For ICT image sequence, two kinds of subvoxel surface detection algorithms： facet model-based method and moment-based method are studied. To solve the problem of large computation of both algorithms, a pre-processing procedure based on Otsu threshold segmentation is proposed, which can reduce the number of surface voxel candidates a lot and accelerate the original one greatly. Finally, we compare the two subvoxel edge detection algorithms in the experiment on aeroengine blade simulation data. The results show that both methods can achieve less than 1/5th pixel accuracy and the introduction of the pre-processing procedure can increase the speed about 4 times.

关 键 词：工业CT 亚体素表面检测 FACET模型 几何矩 

分 类 号：TP274.4[自动化与计算机技术—检测技术与自动化装置]