The feasibility of automated identification of six algae types using neural networks and fluorescence-based spectral-morphological features