摘要
线图匹配在计算机视觉领域有着广泛的应用。该文研究基于改进的NRLCC编码以及面之间的关系进行立体线图的匹配。首先利用图同构检测目标线图与模型线图的面的排列关系是否一致;然后获取面的NRLCC编码和ENRLCC编码,分别得到面的顶点的凹凸性信息和组成面的曲线段的凹凸性信息。最后,利用以上几组信息,进行线图的匹配。试验结果表明,该方法可以对具有曲线边缘的工业物体进行准确的识别。
Line drawing matching is used widely in computer vision field.The paper researches how to match line drawings using improved NRLCC code and relation of faces.First,the array relation of objective line drawing's face and model line drawing's face should be tested using graph isomorphism;Second,we get the NRLCC code and ENRLCC code.They describe the concavo-convex information of a face's vertex and the concavo-convex information of a face's curve;Finally,matching can be executed using these informations.Test result make it clear that this algorithm can recognize manufactured objects with curve edges well and truly.
出处
《计算机工程与应用》
CSCD
北大核心
2006年第21期35-37,42,共4页
Computer Engineering and Applications
基金
国家自然科学基金资助项目(编号:60275037)
江西省自然科学基金资助项目(编号:0311019)
江西省测试技术与控制工程研究中心开放基金资助项目(编号:2002.007)
关键词
线图
同构
NRLCC编码
匹配
line drawing,isomorphism,NRLCC code,matching