Building detection using airborne full-waveform LiDAR data

Abstract

Different from the traditional discrete airborne laser scanning (ALS) systems, the latest airborne full-waveform laser scanning is capable of recording complete waveform of backscattering pulses. Through this property, it becomes more possible to detect objects lying on the path of laser pulse and capture additional information about targets, and therefore such system has been applied to detect objects covered by thick forests or vegetation. In this study we focus on extracting buildings from laser point clouds in forest areas. To this purpose, we firstly indicated possible position of buildings from first echo data. Waveform indicators were subsequently applied to classify points of buildings from all points. The waveform indicators we took into account were the amplitude, the echo width and the backscatter cross-section. At last, a classification strategy using the information derived from full-waveform LiDAR data was developed to distinguish building and vegetation points. The overall building points were derived.