An enhance ranging algorithm based on multi-waveform classification with hyperspectral LiDAR

Abstract

When the light detection and ranging (LiDAR) detected system interrogates targets with high retro-reflective properties, the backscattered pulse energy is prone to exceed the maximum receiving range of the photodetector. It will cause signal saturation because the system cannot record the complete echo waveform, which leads to the deviation of ranging results. To solve the problem, this study proposes an enhanced ranging algorithm based on multi-waveform classification (RMWC) with 101-band hyperspectral LiDAR (HSL). Considering the difference of echo waveform collected in different bands, the threshold method is combined with random forest classifier to classify echo waveforms into normal waveform, saturated waveform and U-shaped waveform. Then peak value method (PV), centroid method (CM) and waveform fitting method (WF) are selected to calculate the corresponding time of flight of the three waveforms respectively. Finally, the target distance is determined by the weighted average of the mode values of the bands’ ranging results. Compared with PV, CM and WF ranging results at 905 nm, the experimental results verify that the proposed algorithm can effectively reduce the ranging error caused by waveform saturation. For planar retro-reflection target, the average error (AE) is 0.0053 m and the standard deviation (Std) is less than 0.0083 m. For diffuse reflection whiteboard, the AE is 0.0010 m and the Std is 0.0053 m. The reconstruction results of planar target point cloud show that RMWC ranging results are better than these traditional single-band ranging methods, which can provide a reference for the optimal design of high precision laser ranging system.