Research on echo waveform identification algorithm of hyperspectral LiDAR

Hyperspectral LiDAR, an innovative active remote sensing technology, captures both spatial and spectral information of targets. When detecting multiple targets with layered distribution, overlap of echo waveforms can occur if the target interval is less than the LiDAR signal pulse width's corresponding distance unit. This paper addresses the challenge of identifying highly overlapping echo waveforms by employing machine learning for waveform classification. The study encompasses the modeling of echo signal modulation for hierarchical target detection, experimental validation, dataset acquisition based on the model, and the application of machine learning techniques. The results indicate that the Random Forest and Support Vector Machine algorithms achieve an 85% classification accuracy for 5cm intervals and an 82% average accuracy for 1-10cm intervals, demonstrating promising prospects for machine learning in classifying LiDAR echo waveforms.