基于特征点的行车数据序列对齐  

Feature based VTDR data alignment

作　　者：江诗锋[1] 何振峰[1] 

机构地区：[1]福州大学数学与计算机科学学院,福建福州350108

出　　处：《福州大学学报（自然科学版）》2010年第2期202-207,共6页Journal of Fuzhou University(Natural Science Edition)

基　　金：福建省青年创新基金资助项目(2006F3075)

摘　　要：依据背景知识,选择行车数据序列在隧道附近所特有的形态,作为该路段的特征点,进行序列的对齐.给定一组未对齐的序列,使用期望最大化(EM)算法进行对齐:在E步算出标准序列,在M步对齐序列.在对齐序列时,采用了基于形态的相似性度量找出待对齐序列与标准序列的最佳特征点匹配方案.本策略成功用于福泉高速行车数据的对齐.A novel feature based alignment technique for passenger vehicle speed series is presented. The feature utilized here is the special outline of speed series near tunnel according to background knowledge. Given a set of series to be aligned （ marked as L′）,the expectation maximization （ EM） method is utilized to help alignment： at the expectation step,construct a standard curve （ SC） . At the maximization step,align all series with standard curve. During the alignment,the best match pair of feature points in SC and L′ was found out according to an outline based similarity measurement. The alignment technique had been applied to the alignment of vehicle traveling data in Fuquan highway successfully.

关 键 词：时间序列 对齐 行车数据 期望最大化(EM) 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]