Automatic assessment of relative accuracy of airborne LiDAR

In this paper an automatic method for assessment of relative accuracy of airborne LiDAR (Light Detection and Ranging) data using surface planes, straight lines and points is presented. Firstly, the LiDAR data is filtered using the progressive morphological filter and RANSAC (Random Sample Consensus) filter. Then, the roof planes are automatically extracted on both strips by region growing algorithm. After, the feature correspondences are established and the transformation parameters are estimated using a variant of ICP algorithm. Finally, in order to evaluate simultaneously planimetric and altimetric relative accuracy, straight lines obtained by intersection of two adjacent surface planes are used. An experiment was conducted to evaluate the proposed method using two overlapping LiDAR strips, whose density is 1 point / m². The assessment of relative accuracy of these data was performed automatically and the results obtained by proposed method, showing values equal -0,0019 cm to mean of the point-to-plane distance and a rmse around 6,04 cm

Keywords:
LiDAR; relative accuracy; Surface planes; Straight lines; ICP