In LiDAR simultaneous localization and mapping (SLAM ),whether the feature points are extracted accurately determines the precision of mapping and positioning in the whole system.Using curvature-based feature point registration cannot fully constrain the pose.Aiming at this problem,a slope-based point cloud preprocessing algorithm is proposed,which can quickly extract line segments(horizontal and vertical line segment)from point cloud.Firstly,the point cloud is serialized in the horizontal direction through the angle,and in the vertical direction through the laser channel number,and the invalid partitions are removed through the partition variance.Then,straight line segments are extracted in the horizontal and vertical directions,and the straight line categories are judged.Finally,the horizontal and vertical line segments are output.Analysis result of experimental data show that,compared with the previous Fast algorithm and Efficient algorithm,the processing speed is up to 150 Hz,and the registration effect is better than using the original point cloud directly.
关键词
地面墙面分割/激光点云预处理/点云特征点提取
Key words
ground and wall segmentation/laser point cloud pretreatment/feature point extraction of point cloud
维普
万方数据