A water toxicity monitoring system based on computer vision technology

Japanese medaka (Oryzias latipes) is highly valuable in the field of monitoring the safety of drinking water. The previous study cannot extract the characteristics which can reflect the toxicity of water real-time and accurately. According to the shortcomings of the previous research, this paper discusses a new water toxicity monitoring system adopting a new observation from the bottom to top, and more effective computer vision algorithms simply. In order to effectively extract the features such as swimming speed changing, gills opening and closing, and pectoral fins and tail swing, we have used the automatic threshold segmentation, foreground extraction, classification, skeleton extraction, morphological and geometrical moment algorithm. The preliminary test results show that the hardware designing and algorithms for extracting the characteristic information of medaka are effective and feasible.