Classification of terrestrial point cloud considering point density and unknown angular resolution

In order to solve the problems of unknown angular resolution and point density variation of terrestrial laser point cloud, a classification method considering density change and unknown angular resolution was proposed in this paper. To improve the traditional point density calculation method, the angular resolution estimation method of random neighborhood analysis was presented. Then we combine angular resolution to propose a grid feature extraction method which takes density variation into account. The proposed method was tested on three datasets. The result shows that the error of our method is smaller than 0.002°, which can accurately estimate the angular resolution. And compared with traditional density feature, our method can improve the overall accuracy of point cloud classification, and perform well in the extraction of cars and pole. The angle resolution can be accurately estimated with this method, and the point cloud can be classified with higher accuracy, which can provide a reference for density adaptive processing of large-scale terrestrial laser point clouds.