摘要
提出一种基于高斯曲率和平均曲率的三维局部相似目标匹配方法.首先筛选出待匹配曲面上固有特征相似的点,形成点对集合.然后利用非对称三角形骨架来定位三维曲面,在点对集合中寻找相似三角形对,并导出其空间变换,构成三角形对集合.最后通过得分函数,求出三角形对集合中空间变换的最佳值,得出最佳匹配.实验表明,该方法对三维局部相似目标匹配具有较好的识别效率,针对不规则三维曲面,能够保证较好的匹配速度.
Based on Gaussian curvature and mean curvature, a 3D partially similar object matching approach is proposed. Firstly, the point-pair set is constructed by filtrating points with similar inherent characteristic. Next, the triangle-pair set is formed after searching similar triangles in the point-pair set. Finally, scoring function is employed to determine the optimal transformation in triangle-pair set. Experimental results show good matching efficiency and running time complexity in the partial surface matching of irregular surfaces.
出处
《模式识别与人工智能》
EI
CSCD
北大核心
2008年第5期586-591,共6页
Pattern Recognition and Artificial Intelligence
基金
国家自然科学基金项目(No.60472061)
国家863计划项目(No.2006AA04Z238)资助
关键词
曲率
微分几何
局部相似目标
匹配
模式识别
Curvature, Differential Geometry, Partially Similar Object, Matching, Pattern Recognition