Laser-Induced Fluorescence Hyperspectral Imaging for Classification of Coastal Macroalgae, Eelgrass, and Mussels

This study explores the application of fluorescence hyperspectral imaging (FHSI) for marine ecosystem monitoring, combining laser-induced fluorescence (LIF) with hyperspectral imaging (HSI), an approach rarely applied in marine ecology. Photopigments such as chlorophylls, carotenoids, and phycobiliproteins emit distinct fluorescence signals when excited at specific wavelengths. Species-discriminating spectral signatures can then address persistent challenges in benthic monitoring, including overlapping optical properties among species and complex underwater conditions. To analyse these subtle interspecific pigmentation differences, we examined their continuous emission spectrum using an inexpensive 450 nm broad-area diode laser as the light source, instead of a broad-spectrum light, in combination with a push-broom hyperspectral camera and a one-dimensional convolutional neural network (1D-CNN). We classified brown, red, and green macroalgae, eelgrass, and blue mussels in laboratory and field experiments in Danish coastal waters, achieving macro-F1 scores of 80.2% and 79.6%, respectively. We show that the method was highly accurate for photopigmented organisms but limited for dark-coloured, low-fluorescence species such as mussels. Field tests identified sunlight and turbidity as the main constraints on in situ performance, while fluorescence fingerprints remained detectable under shaded conditions. These findings establish FHSI as a promising tool for automated benthic habitat analysis and remote-sensing-based marine ecosystem monitoring.