Recognition of colonial and multicellular microplankton organisms in surface waters using computer vision

When assessing the quality of surface water by biological indicators, there is a problem of measuring the concentration of microplankton by individual classification groups, and this involves the recognition of microorganisms. A method is proposed for increasing the accuracy of an information-measuring system that measures the concentration of microplankton in a running stream. The article considers the problem of accuracy increasing in recognition and counting of colonial and multicellular microplankton organisms in determining the biological indicators of water quality in natural reservoirs using computer vision. In the paper was presented a method has been developed to improve the recognition accuracy of colonial and multicellular organisms of microplankton, which is notable for low computational complexity and allows increasing the accuracy of measuring the concentrations of various groups of microplankton. Experimental results confirmed that the use of the developed modified algorithm allows to increase the accuracy of recognition of colonial algae Pediastrum duplex and multicellular Ulothrix and reduce the number of false-positive recognition of them as several unicellular organisms.