An Average Error Ellipsoid Model for Evaluating Precision of Point Cloud from TLS

Abstract

A terrestrial laser scanner (TLS) measures individual point with a precision in the order of millimeters. The evaluation of point cloud precision, although directly affecting the application of TLS like the precision of deformation extraction and digital elevation model (DEM) reconstruction, is still not well understood. Point cloud precision is different from individual point precision, which was influenced by positional error and the adjacent error spaces. This paper is focused on a new evaluation model of point cloud precision based on the point cloud error ellipsoid. The kernel of this model is the computation of the overlap of adjacent error ellipsoid and the determination of functional relationship between average error ellipsoid volume and point cloud accuracy. Furthermore, the effectiveness of the evaluation model of point cloud precision is discussed with a validation experiment. This paper briefly outlines the key advantage of the proposed evaluation model, such as the capability of providing local point cloud precision of building or other architecture approximately plane.