Application of Wavelets and Kernel Methods to Detection and Extraction of Behaviours of Freshwater Mussels
Some species of mussels are well-known bioindicators and may be used to create a Biological Early Warning System. Such systems use long-term observations of mussels activity for monitoring purposes. Yet, many of these systems are based on statistical methods and do not use all the potential that stays behind the data derived from the observations. In the paper we propose an algorithm based on wavelets and kernel methods to detect behaviour events in the collected data. We present our algorithm together with a discussion on the influence of various parameters on the received results. The study describes obtaining and pre-processing raw data and a feature extraction algorithm. Other papers which applied mathematical apparatus to Biological Early Warning Systems used much simpler methods and their effectiveness was questionable. We verify the results using a system with prepared tags for specified events. This leads us to a classification of these events and creating a Dreissena polymorpha behaviour dictionary and a Biological Early Warning System. Results from preliminary experiments show, that such a formulation of the problem, allows extracting relevant information from a given signal and yields an effective solution of the considered problem.
KeywordsAutomated biomonitoring Biological Early Warning System Wavelets Time series Zebra mussel (Dreissena polymorpha)
Unable to display preview. Download preview PDF.
- 2.van Hoof, F., Sluyts, H., Paulussen, J., Berckmans, D., Bloemen, H.: Evaluation of a bio-monitor based on the phototactic behavior of daphnia magna using infrared detection and digital image processing. Water Sci. Technol. 30, 79–86 (1994)Google Scholar
- 8.Kramer, K.J.M., Botterweg, J.: Aquatic biological early warning systems: an overview. In: Jeffrey, D.J., Madden, B. (eds.) Bioindicators and Environmental Management, pp. 95–126. Academic Press, London (1991)Google Scholar
- 9.Wiśniewski, R.: New methods for recording activity pattern of bivalves: A preliminary report on dreissena polymorpha pallas during ecological stress. In: Tenth Intern. Malacol. Congress, pp. 363–365 (1991)Google Scholar
- 14.Rutkowski, L.: Adaptive Filters and Adaptive Signal Processing (in Polish). WNT, Warsaw (1994)Google Scholar