基于激光点云特征匹配的物流分拣目标实时跟踪方法

Real time tracking method of logistics sorting target based on laser point cloud feature matching

为准确定位物流信息,降低货物丢失概率,提出了基于激光点云特征匹配的物流分拣目标实时跟踪方法。通过激光三维扫描采集点云数据,经过奇异值分解与坐标转换实现孤立点云数据合并,利用中值滤波算法进行噪声过滤,通过对图像帧的分割处理采集目标边缘轮廓;将全局与局部轮廓特征匹配方式计算轮廓点数量与特征向量;分别获取物流分拣目标跟踪中心的测量值与预测值,综合考虑匹配置信度、测量值和预测值之间的约束关系,确定目标跟踪位置。仿真分析表明,该方法能有效去除点云数据噪声,且跟踪轨迹与实际轨迹吻合度较高。

In order to accurately locate the logistics information and reduce the probability of goods loss, a real-time tracking method of logistics sorting target based on laser point cloud feature matching is proposed. The point cloud data is collected by laser three-dimensional scanning, and the isolated point cloud data is merged by singular value decomposition and coordinate transformation. The noise is filtered by median filtering algorithm, and the target edge contour is collected by image frame segmentation;The global and local contour features are matched to calculate the number of contour points and feature vectors;The measured and predicted values of the logistics sorting target tracking center are obtained respectively, and the target tracking position is determined by considering the matching reliability and the constraint relationship between the measured and predicted values. Simulation results show that the method can effectively remove the noise of point cloud data, and the tracking trajectory is in good agreement with the actual trajectory.