基于参数统计的LiDAR数据分割算法

LiDAR Segmentation Based on Statistics Analysis

假设自然地表分布连续,在地形起伏不大的情况下,局部地表的高程值具有正态分布的特点,而地物点的存在往往会破坏正态分布的平衡性。基于这种思想文中引入偏度和峰度两个统计量,设计了基于参数统计的LiDAR分割算法。该方法无需事先设定阈值,实施简单方便,不仅适用于格网形式LiDAR数据,也可用于离散分布的激光脚点。

Assumed that the nature surface is continuous and sloping gently, the height values in local area have the characteristics of normal distribution. The existed non ground points affect the balance of normal. To separate all ground and object points, two statistics were introduced here, respectively skewness and kurtosis. A new LiDAR segmentation method based on statistics analysis was proposed based on skewness and kurtosis balance. The method is simple and effective, not only fit for grid data, but for irregular ones.