摘要
提出一种从激光扫描数据中提取角点特征的方法,以满足结构化环境中移动机器人定位的需要。该方法分为角点特征滤波、角点函数计算、非极大值抑制和角点特征提取与定位四部分。首先,通过角点特征滤波排除不具备角点特征要素的扫描点。然后,通过角点函数衡量余下的扫描点作为角点特征的程度。之后,采用非极大值抑制方法去除局部区域内非极大值对角点特征提取的干扰。最后,通过阈值法提取出角点特征并对其进行定位。实验结果表明,该方法可以准确地从激光扫描数据中提取出角点特征,并对位置参数具有很好的鲁棒性。
A method is presented for extracting corner features from laser scan data for mobile robot localization. The method includes corner feature filter, corner function calculation, non-maxima suppression, and corner feature extraction and localization. First, the scan point without corner feature elements is excluded from all scan points by corner feature filter. Second, corner function is calculated to measure the extent that the remained scan points are corner features. Third, non-maxima suppression method is used to eliminate disturbance resulted from non-maxima in local area in extracting corner features. Finally, corner features are extracted by threshold method and localized. Experiment results show that the method can extract corner feature accurately from laser scan data and is robust to position parameter.
出处
《南京航空航天大学学报》
EI
CAS
CSCD
北大核心
2012年第3期379-383,共5页
Journal of Nanjing University of Aeronautics & Astronautics
基金
江苏省科技支撑计划(BE2010189)资助项目
关键词
移动机器人
定位
特征提取
激光雷达
mobile robots
localization
feature extraction
laser radar