激光与光电子学进展, 2020, 57 (6): 061012, 网络出版: 2020-03-06
基于激光点云全局特征匹配处理的目标跟踪算法 下载: 1639次
Object Tracking Algorithm Based on Global Feature Matching Processing of Laser Point Cloud
图像处理 目标跟踪 激光点云 目标识别 三维全局特征 激光雷达 image processing object tracking laser point cloud object recognition three-dimensional global feature lidar
摘要
实际场景中各物体的尺寸差异导致激光三维数据中各物体对应的三维积分区域(SVR)存在差异。在初始帧中,借助于SVR筛选与全局特征匹配完成目标识别,实现对待跟踪目标的自动选取,并且比较四种全局特征描述子的识别能力及运行速度。得到初始帧中的目标位置后,提出了利用全局特征匹配在后续帧中实施目标跟踪的方法。实验结果表明,SVR筛选有利于提高识别跟踪的准确率及算法整体运行速度。
Abstract
The difference in size between different kinds of objects will lead to the difference in summed volume region (SVR) of corresponding laser point cloud. In the first frame, object recognition is accomplished based on SVR selection and global feature matching to automatically select the interested object. The performance and execution time of four global feature descriptors are compared. After obtaining the position of the interested object in the first frame, an object tracking method based on global feature matching processing of laser point cloud is put forward for subsequent frames. The experimental results show that adding SVR selection is helpful to improve the accuracy of recognition and tracking and the overall running speed of the algorithm.
钱其姝, 胡以华, 赵楠翔, 李敏乐, 邵福才. 基于激光点云全局特征匹配处理的目标跟踪算法[J]. 激光与光电子学进展, 2020, 57(6): 061012. Qishu Qian, Yihua Hu, Nanxiang Zhao, Minle Li, Fucai Shao. Object Tracking Algorithm Based on Global Feature Matching Processing of Laser Point Cloud[J]. Laser & Optoelectronics Progress, 2020, 57(6): 061012.