A method for separation of the terrain and non-terrain from Vehicle-borne Laser Scanning Data

Half the points from vehicle-borne laser scanning data are terrain data. If you want to extract features such as trees, street lights and buildings, terrain points must be removed. Nowadays, either airborne or vehicle-borne laser data, are mostly used to set an elevation threshold based on the scanning line or POS data to determine whether the point is a terrain point or not, but the disadvantage is part of low buildings or other feature objects will be lost. If the study area has high differences in the horizontal or the forward direction, this method is not applicable. This paper investigates a new methodology to extract the terrain points, which has great significance for data reduction and classification. The procedure includes the following steps: 1)Pre-processing: to remove discrete points and abnormal points. 2) Divided all the points into grid, calculating the average value of the XY and the minimum value of the Z of all the points in the same grid as the central point of the grid.3) Choose nearest six points which are close to the centre point to fitting the quadratic surface.4)Compare the normal vector of the fitting surface of the grid to the normal vector of the 8-neighborhood, if the difference is too big, it will be smoothed.5) Determine whether the point in the grid is on the surface, if the point belongs to the surface, it will be classified as terrain point. The results and evaluation have shown the effectiveness of the method and its potential in separation of the terrain of various areas