Terrestrial Lidar Data Classification Based on Raw Waveform Samples Versus Online Waveform Attributes

Classification of Terrestrial Lidar Data Directly From Digitized Echo Waveforms

“…The contributions of this study are two-fold: 1) regarding the promising results achieved in the earlier study [37], the utility of the digitized waveform samples (DNs) collected by a FW TLS system is explored for direct echo (point) classification in a multiple-echo waveform scenario rather than limiting to only single-echo waveforms; 2) calibrated waveform features for each echo, derived from online waveform processing, are examined for multi-class point (echo) classification and classification performance is compared with the performance achieved through classification of the feature vectors constructed from the raw digitized waveform samples of the corresponding echo.…”

“…This paper investigates the utility of the raw digitized waveforms recorded by a FW TLS system for multi-class point cloud classification. The main goal is to extend an earlier experiment, carried out by the same authors [37], that directly exploited the waveform for point cloud classification. Earlier experiments demonstrated the usefulness of exploiting digital samples of recorded echoes for direct classification of lidar points corresponding to single-echo waveforms.…”

“…In this experiment, the same DCNN architecture proposed in the earlier work, accomplished by the same authors [37], is employed for point cloud (echo waveform) classification. The DCNN architecture is shown in Fig.…”

“…Although a large number of studies have focused on different FWA approaches and derived waveform attributes from the modeled echo waveform to characterize target properties, the information content of the raw waveform signal, without apriori decomposition into target atrributes, has not been explored. This is more critical in FW TLS systems, where the traditional FWA approaches, similar to those applied on FW data acquired by airborne lidar systems, may not be applicable [35]- [37].…”

End-to-End Full-Waveform Echo Decomposition Based on Self-Attention Classification and U-Net Decomposition

“…Also, the experimental results showed that, Point2wave is able to translate the 3-D point clouds into desired waveform signals, and the translated waveform signals achieved nearly the same classification performance as the real waveforms. And in paper [39], the authors proposed a DCNN-based classifier for waveform feature classification. The potential of raw samples from full-waveform terrestrial laser scanning systems were explored for point cloud classification in city and countryside environments.…”