Multispectral LiDAR-based underwater ore classification using a tunable laser source

Abstract

Light Detection and Ranging (LiDAR) technology has been extensively used to collect precise geometric information. However, conventional LiDAR sensors are restricted to single-wavelength operation, limiting their ability to capture valuable spectral information. Multispectral LiDAR (MSL) systems, on the other hand, have emerged as a promising solution by facilitating the active acquisition of spatial and spectral data. These advanced systems exhibit tremendous potential in facilitating advanced analysis and classification of seafloor sediments. In this study, the first demonstration of an MSL system optimized for investigating the feasibility of underwater ore classification using a tunable laser source is described. The MSL prototype features a spectral resolution of 10 nm, and 11 spectral channels, covering the range from 460 to 560 nm. Laboratory-based experiments were conducted to evaluate the accuracy of range measurements and the classification performance of the system. The spectral profiles of eight distinct ore samples acquired by the MSL were utilized for classification using four classification methods: naive Bayes (NB), k Nearest Neighbor (KNN), support vector machine (SVM), and random forest (RF). Furthermore, comparative analyses were conducted to investigate the classification enhancements realized by leveraging the MSL system with multiple spectral channels as opposed to single-wavelength and dual-wavelength systems.