激光测量数据特征点识别方法探讨  

Discussion on the Method for Identifying Feature Points from Laser Measurement Data

作　　者：李澍[1] 冯国卫[1] 

机构地区：[1]上海交通大学模具CAD国家工程研究中心,上海200030

出　　处：《应用激光》2014年第5期451-454,共4页Applied Laser

摘　　要：分析了激光测量数据的分布特点,并据此介绍了有效数据的筛选方法和数据滤波的方法。重点探讨了利用曲率计算获得特征点的方法并提出了利用多项式拟合识别特征点的方法,通过算例分析验证了两种方法的优缺点和适用范围,结果显示在激光测量数据分布比较平坦的情况下,多项式拟合算法要优于曲率计算方法。研究可为不同工业应用领域激光采集数据后选取特征点计算方法提供参考。The distribution characteristic of points obtained by laser measurement has been analyzed in this work.Based on the distribution of the data,the methods for screening the valid data and data filtering are introduced.Two technics for identifying feature points are exploited,which are the classical curvature calculation and polynomial fitting.Experiments have been executed to verify and validate the technics.The advantages and limitations of the technics for identifying feature points are analyzed.The results show that the identified feature point by polynomial fitting method is more accurate than that by curvature method.The study can offer some references for selecting different methods for the feature points＇ identification from laser measurement data in different industrial application area.

关 键 词：测量 数据处理 特征点 曲率计算 多项式拟合 

分 类 号：TN249[电子电信—物理电子学]