Algae classification and identification based on snapshot spectral microscopy imaging technology

Algae and algae communities serve as indicator species in ecosystems, reflecting water body conditions and pollution levels. Accurate and efficient species identification of algae facilitates environmental monitoring and ecological research. This study proposes an algae classification method integrating spectral microscopy imaging technology with neural network algorithms. It achieves rapid identification of five algae species-Anabaenaazotica, Uronemaelongatum, Chlorella, Dunaliellasalina, and Haematococcuspluvialis-by acquiring spectral images of different algae and analyzing their absorption spectral characteristics. Separate ResNet50 convolutional neural network classification models were constructed using grayscale images and spectral images of characteristic bands. Experimental results demonstrate that the classification model based on feature spectral images achieves an accuracy of 97.14% and a recall rate of 96.67% on the test set, significantly outperforming classification based on grayscale images. This study validates the application potential of snapshot spectral microscopy in algae identification, providing a simple, rapid, and low-cost detection method for algae classification.