VALIDATION OF LIDAR SURVEY DATA BY COMPARISON OF SEVERAL UNCERTAINTY MODELS
This paper introduce a new method for validating the precision of an airborne or a mobile LiDAR data set. The proposed method is based on the knowledge of an a Combined Standard Measurement Uncertainty (CSMU) model which describes LiDAR point covariance matrix and thus uncertainty ellipsoid. The model we consider includes timing errors and most importantly the incidence of the LiDAR beam. After describing the relationship between the beam incidence and other variable uncertainty (especially attitude uncertainty), we show that we can construct a CSMU model giving the covariance of each oint as a function of the relative geometry between the LiDAR beam and the point normal. The validation method we propose consist in comparing the CSMU model (predictive a priori uncertainty) t the Standard Deviation Alog the Surface Normal (SDASN), for all set of quasi planr segments of the point cloud. Whenever the a posteriori (i.e; observed by the SDASN) level of uncertainty is greater than a priori (i.e; expected) level of uncertainty, the point fails the validation test. We illustrate this approach on a dataset acquired by a Microdrones mdLiDAR1000 system.