一种改进SIFT算法的单目视觉里程计  

An Improved-SIFT Algorithm for a Monocular Visual Odometry

作　　者：赵黎明[1] 陈宁[1] 

机构地区：[1]集美大学机械与能源工程学院,福建厦门361021

出　　处：《集美大学学报（自然科学版）》2015年第3期218-224,共7页Journal of Jimei University：Natural Science

基　　金：福建省科技厅省属高校专项课题资助项目(JIC2014024)

摘　　要：针对如何准确获取位姿信息来实现移动机器人的避障问题,提出一种改进SIFT特征点匹配的单目车载视觉里程计算法.首先,为了提高特征点匹配的正确率和实时性,结合主成分分析法和平面极线几何约束,改进了传统SIFI匹配算法,其次,建立合理的移动机器人运动数学模型,得到连续帧间图像信息和移动机器人运动位姿变化的转换关系.试验结果表明,误差仅为1.6%,算法运行时间缩短0.022 s.This paper presents a robust method for monocular visual odometry capable of accurate position estimation of a mobile robot based on an improved SIFT algorithm. Firstly, for improving the accuracy and performance of real-time of feature points matching, improvements have been made on SIFT algorithm by combining Principal Component Analysis and Epipolar geometric constraint. Secondly, a mathematical model has been developed to determine the relationship between the information of consecutive frames and pose change of mobile robot. Results show that the algorithm brought about an accuracy with relative error less than 1.6%, and the running time of the algorithm reduces 0. 022 seconds correspondingly.

关 键 词：视觉里程计 改进SIFT 平面极线几何约束 

分 类 号：TP24[自动化与计算机技术—检测技术与自动化装置]