工业CT图像轮廓数据提取研究  

Extraction of contour data based on industrial CT image

作　　者：曹冰[1] 何朝明[2] 李柏林[2] 

机构地区：[1]四川工程职业技术学院,德阳618000 [2]西南交通大学机械工程学院,成都610031

出　　处：《现代制造工程》2011年第1期82-84,127,共4页Modern Manufacturing Engineering

基　　金：德阳市重点科技攻关项目(2006ZG003)

摘　　要：零件CT扫描设备在扫描过程中不可避免会产生大量冗余信息,导致获取的零件轮廓离散点数据存在噪声和冗余数据,使得特征点提取变得异常困难。传统轮廓数据点的提取和平滑除噪是采用不同方法分开进行的,这种传统方法可能会剔除有用的特征数据和造成得到的边界点数据发生畸变和偏移。利用多尺度分析技术研究轮廓离散数据的处理,通过对原始轮廓离散数据曲率进行小波多分辨率分解,然后根据尺度间相关性来去除轮廓数据冗余信息和提取特征点,便于在随后的重建过程中实现更精确的表达。基于该方法开发的软件,在实际应用中取得了良好的效果。During the CT scanning process for parts, it will inevitably produce a large number of useless redundant information, resuiting in existence of noise and redundant data of contour data points of parts, making extraction of characteristics become extremely difficult. Traditionally, the use of different methodology of extraction and smoothing the noise of contour data points is car- ried out separately, which may remove some useful data and even give rise to distortion of boundary points data. The use of muhiscale analysis technology to process the contour data, which decomposes the curvature of original data via wavelet multi-resolution, is helpful in the subsequent reconstruction process to achieve a more accurate expression based on relevant redundant data a- nalysis and characteristics. The software on the basis of this methodology has achieved good results during its application.

关 键 词：工业CT 多尺度变换 小波分析 轮廓数据提取 

分 类 号：TH123[机械工程—机械设计及理论]