摘要
给出了一种结合最小有向包围盒(minimum oriented bounding box,MOBB)以及形状分布的识别圆台体及其参数的算法。首先建立基本体素的MOBB,对体素进行标准化;然后生成其D2形状分布曲线,计算此曲线分别与标准圆锥体形状分布曲线、标准圆柱体形状分布曲线的EMD(earthmover’s distance)值以及相应的上底直径;最后通过比较两个上底直径的大小来判断此体素是否为圆台体,并由包围盒的边长给出圆台体的参数。该算法简单有效,不需要任何交互操作,而且能够识别发生了一定外形改变的圆台体,抗噪能力强。
Frustum of cone and its parameter recognition algorltnm based on shape distributions and minimum oriented bounding box was given. Firstly, we normalized the basic voxel by construc- ting its minimum oriented bounding box. Secondly, we constructed its D2 shape distribution curve, and computed the values of EMD（earth mover＇s distance） compared with the corresponding curves of the normal cone and cylinder respectively, and the corresponding top diameters. Finally, we compared the top diameters and obtained the parameters by the bounding box. This algorithm needs not interactive manipulation and has a strong antinoise ability to recognize the basic voxel with small changes of shape.
出处
《中国机械工程》
EI
CAS
CSCD
北大核心
2009年第9期1102-1106,共5页
China Mechanical Engineering
基金
国家自然科学基金资助项目(60573177)
航空基金资助项目(04H3059)
关键词
逆向工程
体素识别
圆台体
形状分布
reverse engineering
feature recognition
frustum of cone
shape distribution
