Radiometric Evaluation of an Airborne Single Photon Lidar Sensor

Abstract

Lidar intensity is correlated with illuminated target physical properties, particularly target reflectance, making it a valuable quantity for applications, such as land cover classification, data registration, structural damage detection, and qualitative point cloud interpretation. In contrast to traditional linear-mode lidar (LML) hardware, single photon lidar (SPL) detectors produce a binary response to impinging photons and therefore do not provide an intensity measure for each detected return. This is a significant drawback but can be addressed by computing a measure of local point cloud density for each point. Since the arrival and detection of single photon are governed by statistics such that the observations of brighter surfaces are more probable to generate a detection event than darker surfaces, a local point cloud density metric can be used as a proxy for traditional LML intensity. We define the relationship between target reflectance and photon detection probability and compare the predicted relationship with empirical observations of ground reflectance and an estimate of detection probability generated from local point cloud density. We also examine a pulsewidth measure provided by the SPL sensor used for this letter, as well as the influence of neighborhood radius on variance in the probability estimates and a filtered version of the hardware-supplied pulsewidth.