Model On Anomaly Behavior Of Fish School For Water Quality Monitoring

Biological monitoring technology makes use of the phenomena that aquatic organisms will change their physiology and behavior to adjust to the contaminated environment, to achieve the aim of water quality deterioration trend forecasting and pollution emergency warning. This technology is more intuitive to reflect the integrated influence and extent of harm on fish behavior under variety of pollutants in the natural conditions, and makes biological monitoring technology one of the most important measure for water quality assessment. Therefore, how to extract bio-behavioral features effectively and build up the relationship between abnormal behavior of bio-indicator and contaminated water are key points in the study of water quality monitoring online.Using fish school as bio-indicator, this paper analyzes the influence of contaminated water on fish school behavior through toxicological experiment and proposes a modeling method of fish shoaling behavior under drug-free water exposed, meanwhile accomplishes a water quality online monitoring system based on this model. The main research works are summarized as follow:First, study the segmentation algorithm of fish school target based on graph theory and uniformity measurement. This efficient graph-based algorithm for fish school target segmentation use the uniformity measurement as image assessment criteria and introduces the Laplacian operator to improves imprecision when dealing with the border and texture and reach the best segmentation result.Second, study the modeling method of fish shoaling behavior. Based on the results of fish school target segmentation, this paper introduces the Delaunay triangulation network algorithm to compute the centroid of fish school and build up the fish shoaling behavior model. By analyzing the statisticial data of fish school centroid and group distribution, this paper extracts the features describing fish shoaling behavior under control.Third, study the online water monitoring method based on fish shoaling behavior model. Based on the result of extracted fish shoaling behavior features, the warming time and warming threshold under contaminated water has been calculated. In the meantime, this paper conducts several online monitoring experiments according to the corresponding contaminants to verify the validity of applying fish shoaling behavior model in online water monitoring.