一种大场景点云的快速分割方法

A FAST SEGMENTATION METHOD FOR LARGE SCENE POINT CLOUD

为实现虚拟展示中大场景三维点云的快速分割,提出了一种基于信号强度统计分布的三维点云分割方法。该方法通过对激光扫描回波信号强度的统计分析,自动获取目标实体信号强度和采样率的最佳阈值关系,再根据阈值实现对无效信号的过滤,能快速提取出场景中的目标实体。实验表明,该方法能取得理想的分割结果,具有较强的实用性。

In order to achieve fast segmentation of three dimensions point cloud of large scene in virtual display, a 3-D point cloud segmen- tation method based on signal intensity statistical distribution is proposed. Through statistical analysis of the signal intensity of laser scanning echo, the method automatically obtains the signal intensity of target object as well as the optimal threshold values relationship of sample-rate. Then, based on the optimal threshold values, it implements the invalid signal filtering and can quickly extract the target objects in scene. Ex- periment results show that this method is able to attain ideal segmentation effect and has preferably strong practicality.