方向自适应的光子计数激光雷达滤波方法

提出一种自适应滤波方向的光子计数激光雷达点云滤波方法,定义了一种可调节主滤波方向的滤波核,通过遍历得到最佳滤波方向的密度值并剔除远离地物的噪声点,根据密度值与邻域内其它点的密度值差值剔除接近地物的噪声点。通过实验数据对算法进行了验证,结果表明算法能有效剔除与地面非常接近的噪声点,适用于低密度地物点云的滤波处理,其中植被滤波精度91.86%,地面点滤波精度97.89%。

An adaptive directional filter for photon counting Lidar point cloud data

An adaptive directional filter method was proposed for the photon counting Lidar point cloud data. The method defines a filter kernel with its main filter direction adjustable. The density of the best filter direction were achieved by traverse and the noise points away from the objects were removed. The noise points adjacent to the objects were eliminated according to the density difference between the point and points in its neighborhood. The filtering method provide here is validated through the point cloud data obtained in an aerial experiment. The results show that the filtering method is able to eliminate the noise points very close to the ground effectively and is fit for the low density object point cloud recognition, while the filter accuracy is 91.86% for vegetable points and 97.89% for ground points.