截面线特征直接提取的离散曲率法  

Direct Feature Extraction of Sectional Curve Based on a Discrete Curvature Arithmetic

作　　者：孟娜[1,2] 周以齐[1] 曲娟[2] 陈新立[2] 

机构地区：[1]山东大学机械工程学院,山东济南250061 [2]山东科技大学机电工程系,山东泰安271000

出　　处：《中国矿业大学学报》2009年第4期571-575,共5页Journal of China University of Mining & Technology

基　　金：山东省自然科学基金项目(Y2006F12)

摘　　要：特征提取是逆向工程的重要步骤,其中截面线特征点的弱化是需要解决的关键问题.提出用离散曲率法提取特征点,曲率表达式包含了高斯核函数曲线,选用了合适的离散尺度因子.根据一阶和二阶离散曲率曲线的局部极值点,确定出截面线特征点集,进行特征点的融合,输出最终的截面线特征点集.通过与文献算法的输出进行比较,表明本算法提取的截面线特征点不容易有漏检和误检,能最大限度地与原有形状特征元保持一致.仿真输出实例证实了本算法的适用性和有效性.The process of feature which the crucial question about points extracted is important weak feature points extracted in the reverse engineering, in of sectional curve needs to be solved. The discrete curvature arithmetic was used to extract feature points, which the curvature expression contains Gaussian nuclear function and chooses appropriate discrete scale factor. Based on the local extremum of one order and two orders discrete curvature, the feature points collection of sectional curve was determined and the feature points was fused. Then the final feature points collection was output. Compared with the feature point output of relevant literature, this method is not prone to slip and miss point features, and can guarantee the consistence between feature extraction results and original model feature element from the experiments. The applicability and validity of this arithmetic was validated through the results of emulation and output of an example.

关 键 词：曲率阶数 离散曲率 截面线 特征提取 高斯核函数 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]