摘要
为实现机械零件的有效分类,针对零件具有比较明显的轮廓特征的特点,提出了基于边界矩的零件图像轮廓特征提取方法。首先采用轮廓提取算法,提取零件二值图像的轮廓,在进行水平倾斜校正后,以零件的质心为中心,将轮廓图像划分为若干个扇形子区域。利用改进的边界矩计算方法,分别计算出各子区域的边界矩,从而得出零件轮廓图像边界矩的分布特性。最后,采用K均值聚类算法对提取的零件轮廓特征进行分类,实验结果证明了该方法的有效性。
To effectively classify mechanical components, a novel method was proposed to extract contour feature of parts image to deal with mechanical parts with comparatively obvious contour features. Firstly, contour extraction algorithm was used to get the contour of mechanical part binary image. After horizontal tilt correction was pro- cessed, the contour image was divided into several fan-shaped sub areas with the centroid as the center. By using im- proved edge moments calculation method, the edge moments for each sub area were calculated. Thus the contribu- tive contour features of mechanical parts were obtained. In the experiments, the K means clustering algorithm was used to classify the parts based on the extracted contour features, and the results proved its effectiveness.
出处
《计算机集成制造系统》
EI
CSCD
北大核心
2008年第7期1375-1379,共5页
Computer Integrated Manufacturing Systems
基金
国家自然科学基金资助项目(59975063)~~
关键词
机械零件图像
轮廓特征
特征提取
边界矩
水平倾斜校正
均值聚类算法
mechanical parts image
contour feature
feature extraction
edge moment
horizontal tilt correction
means clustering algorithm
