Design And Research Of Water Quality Bio-Monitoring Sensor Based On Dynamic Image Understanding

To solve water pollution problem radically, prevention should go before water-control. Thus, water pollution early warning was of great importance. Aquatic animals were usually used as water quality bio-indicators, and water pollution could be detected by their biological conditions early. The paper designed a water quality bio-monitoring sensor based on dynamic image understanding, which integrated technologies of biological water quality monitoring, computer vision, dynamic image understanding, embedded system and etc.. It provided a new method for on-line water quality bio-monitoring and water pollution early warning.The main work in this paper was summarized as follows:1. A water quality bio-monitoring technology was proposed, which was based on dynamic image understanding. Computer vision was used to observe aquatic animals continuously, which were bred in the water quality bio-monitoring sensor designed in the paper. And dynamic image understanding technology was used to analyse their biological conditions to assess water quality, so as to provide water pollution early warning.2. A water quality bio-monitoring sensor was designed, which was miniaturization and intelligentized and could easily be set in a large-scale water field. By its special structure, the aquatic animals of different biological conditions used as water quality bio-indicators could be separated automatically. Computer vision and embedded system were set in the sensor to capture video images of aquatic animals in the sensor and to analyse their biological conditions to assess water quality on-line.3. Image semantics was used to describe the course of water quality bio-monitoring using fish as water quality bio-indicator. Changes of water quality could be indicated by fish survival and activity rates and their abnormal behaviors. These biological conditions of fish could be identified and analysed with image semantics. Dynamic image understanding technology was used to analyse fish's color and movement features, and the foreground objects were detected to assess fish's biological conditions and water quality by analyzing the objects' image semantics and behavior semantics.4. Fish were adopted as water quality bio-indicators in experiments. Fish's biological conditions in fresh water, polluted water and water of different polluted levels were analysed and compared using dynamic image understangding technology. The experiments' results showed that water quality could be reflected by fish's biological conditions, and the indexes of biological conditions could be analysed by dynamic image understanding technology rapidly and effectively. And water quality could be judged by these indexes.The water quality monitoring technology proposed in this paper could be applied for water quality monitoring in rivers, lakes and water factories and etc., which had a broad application foreground. With different water quality bio-indicators and in different water environments, there were still lots of research to do.