Assessment of lidargrammetry for spatial data extraction

Lidargrammetry concerns the production of inferred stereopairs (ISPs) from LiDAR intensity images, intended to stereodigitize spatial data in digital photogrammetric stations. The production of ISPs is based on the principle of stereoorthomates and the extraction of a derivative intensity image, in which an artificial x-parallax is being introduced; other techniques have also developed in order to best utilize the 3D nature of LiDAR data. Lidargrammetry is a relatively new approach, not yet assessed properly, in order to quantify its derivative spatial data quality and the impact of its reduced photointerpretative ability, comparing to typical photogrammetric stereomodels. In this paper, a dense point cloud of 55 points/m2 is used, which is thinned out to 25 points/m2 and 7 points/m2 in order to simulate scan missions of lower pulse repetition frequency. ISPs are being produced from each of these point cloud’s intensity images using the slope parallel projection method and building footprints are being extracted. Using the denser point cloud’s footprints as control data, the relative accuracy of the thinner point cloud’s footprints is assessed, in order to evaluate the effect of the decreasing resolution in the digitization process. Estimated footprint’s relative accuracy (2σ) is 0.5m and 1m for the 25 points/m2 and the 7 points/m2 clouds respectively. Moreover, a reference footprint dataset was derived, by a stereorestitution procedure, using high resolution optical aerial images. Absolute spatial accuracy ranges around 1.5 m making the Lidargrammetric technique capable for extracting spatial data suitable even for large scale mapping.