基于边缘检测滤波的机载激光雷达点云数据处理

Data Processing of Airborne Lidar Point Cloud Based on the Edge Detection Filtering Algorithm

机载激光雷达(Lidar)是一种主动式对地观测技术,可以直接获取点的三维坐标。Lidar点云数据的括滤波和分类,是Lidar数据处理的重要步骤。利用国际摄影测量与遥感协会ISPRS提供的实验数据,采用边缘检测滤波算法和线性卷积滤波算法对数据进行滤波,滤波后的图像表明,边缘检测滤波算法效果优于线性卷积滤波。采用基于Axelsson的改进的不规则三角格网加密方法进行点云分类,将Lidar点云分为以下8类:低点、孤立点、空中点、地面点、模型关键点、低于地表的点、建筑物点和植被点。分类后的Lidar点云数据都被分到了唯一的类别中,清楚地显示出地面信息。结果表明,采用的滤波和分类算法有效可行,对Lidar点云数据处理有重要的借鉴意义。

As an active earth observation technology, Airborne Light Detection and Ranging （Lidar） can acquire the three di- mensional coordinates of points directly. The data processing of Lidar point clouds includes filtering and classification, which is a key step in Lidar data processing. The dataset is obtained from International Society for Photogrammetry and Remote Sensing （ISPRS）. The edge detection filtering algorithm and the linear convolution filtering algorithm were used to filter the data, and results indicated that the filtered image with edge detection filtering algorithm has a better effect than those with linear convolution filtering algorithm. Based on Axelsson＇s modified encrypting Triangulated Irregular Network, Lidar point clouds are classified into eight categories： low points, isolated points, air points, ground, model keypoints, below surface, buildings and vegetation. All points are classified to a unique category with this algorithm, with ground details displayed clearly.