Research And Application On Water Quality Bio-monitoring System Based On Dynamic Image Understanding

With the rapid economic development, China's water pollution has become quite serious, which seriously impacted on people's lives and economic sustainable development. Therefore, we must carry out monitoring of water quality as soon as possible, not "treatment after pollution." Living creatures in the water is closely related to water, they can response changes in water quality through its changes of ecological conditions, and then provide people with the early pollution of water in time. The essence of Bio-style water quality monitoring is passing through changes of ecological conditions of biological indicators in the water, to achieve the purpose of comprehensive evaluation of water quality. This paper presents a technique of Live-fish water quality monitoring based on dynamic image understanding. It combines the water quality biological monitoring, computer vision and image understanding technologies and so on, and provides a new technology and method for the exploitation of Bio-style water quality real-time online monitoring.The main work in the paper includes the following:1. The paper raises a technique of Bio-style water quality monitoring based on dynamic image understanding. The technique adopts computer vision technology to obtain the video image of living creature in Bio-style water quality monitoring as for biological indicator, and then tracks the moving objects of the video serial images to extract the behavioral parameters, and then makes uses of dynamic image understanding technology to analysis these behavioral parameters. What follows next is to achieve the purpose of evaluation of water security condition.2. The paper researches the technology of the video image processing. The study is some pre-processing and processing techniques of video image. To the pre-processing technology of video image, this paper mainly introduce the typical image filtering,smoothing techniques and enhancing technique, and then use these techniques to process video images as the basis for subsequent processing. To the processing technology of video image, this paper mainly introduces the methods of several moving object detection and track. The paper mainly adopt a typical background reconstruction and upgrading technique based on mixed Gaussian model, a technique of object detection based on background subtraction and a technique of object track based on Kalman filter to realize the behavior trajectory tracking of Live-fish.3. The paper executes the semantic description and interpretation to the process of Live-fish monitoring the water quality changes. The swimming speed, swimming height and living number and other behavioral characteristics can reflect the changes of water, and these behavioral characteristics of Live-fish can also can be identified and analyzed by high-level image semantic. This paper designs and extracts the behavioral parameters of Live-fish by tracking the trajectory of fish, and executes semantic analysis to these extracted behavioral parameters, which can judge the ecological conditions of Live-fish.4. Using Live-fish as the biological indicator of Bio-style water quality monitoring to carry out experiment. There have some obvious advantage of fish as water biological indicator, such as low cost, easy access resources and high sensitivity to the water feeling and so on, which makes fish to be the first choice for our experiment. In the experiment, the paper adopts the technique of dynamic image understanding to analysis the ecological conditions and behavioral changes of fish in the normal water, mild and severe polluted water. The result of experiment show that the ecological conditions of fish can reflect the security conditions of water.The technique of Bio-style water quality monitoring has broad application prospects, but because of the limitations of this experiment, the content study of this paper is limited, hoping to provide a valuable reference for future research.