Automatic methods for the detection of harmful algae

The detection of harmful algae is obviously of great economic importance to leisure, tourism and fisheries world–wide. Yet there are no robust techniques that may be deployed in the field for the routine assay of harmful algae in water samples. A number of seemingly promising avenues have been explored in recent years from HPLC to flow cytometry, working on the detection of specific poisons, or other characteristic chemical signatures and even on the morphology of the algae. But none are resilient to the natural variation found in parameters under scrutiny. The recent application of advanced pattern detection algorithms to this difficult problem provides a glimmer of hope in the successful detection by automatic means of harmful algae. This paper will present recent progress in the application of neural network techniques to the recognition of algae and other marine organisms to current techniques.