摘要
移动机器人在未知环境中利用视觉信息进行同时定位与地图构建时,如何提高机器人路标观测质量,成为了视觉SLAM研究中的重要问题。文中首先采用SIFT算法进行视觉图像信息特征点的提取和匹配,然后结合一点随机采样一致性原理,使用扩展卡尔曼滤波算法进行视觉路标的预测和更新。实验结果表明,与传统的基于FAST角点检测的路标观测方法相比,本文提出的SIFT-1pRANSAC-EKF单目视觉SLAM路标观测方法能够有效提高视觉路标的匹配成功率,提高观测质量。
It is an important issue that how to improve the quality of observation in the study on visual SLAM when mobile robots use visual information for simultaneous localization and mapping in an unknown environment. Firstly,this paper uses SIFT algorithm for image feature point extraction and matching,then one point random sample consensus is combined,last it uses the extended Kalman filter algorithm for forecasting and updating of visual landmarks. The experimental results show that observation method of visual landmarks named SIFT-1pRANSAC-EKF in monocular SLAM can improve the success rate of matching landmarks compared with method based on FAST corner detection algorithm,and it improves the quality of observation.
出处
《信息技术》
2017年第3期1-4,8,共5页
Information Technology
基金
国家自然科学基金资助项目(61203365)
江苏省自然科学基金资助项目(BK2012149)
江苏省研究生创新项目(KYLX15_0496)