The Video Image Understanding Of Live Fish Used For Water Quality Monitoring

To solve the increasingly serious water pollution problem, the most effective means is to strengthen water quality monitoring. Thus, the study of water quality monitoring technology is quite important. Following the traditional physical and chemical testing technology of water quality, water quality monitoring which is carried out by application of aquatic organisms has been widely studied, formed water quality bio-monitoring technology in recent decades. The essential of water quality bio-monitoring is evaluating the change of water quality by the change of ecological condition and behavioral habits of aquatic organisms living in the water. This paper combines water quality bio-monitoring technology and video image understanding technology, proposes a thought of video image understanding, which is used in water quality monitoring, it supplies a new technology for the development of water quality bio-monitoring system which is real-time and on-line.This paper carries out particular research on the key technologies of living fish video images understanding of water quality monitor, the main technologies contain preprocessing techniques such as image denoising, smoothing and enhancement, etc, and processing techniques such as moving object detection, background reconstruction, moving object track, etc. In addition, according to the experiment requirements, this paper designs and extracts behavior parameters of testing organism and establishes mapping relationship of testing organism behavior semantics and water quality condition semantics by tracking trajectories of the objects. In this way, it achieves the purpose of evaluating the water quality by understanding the behavior of the testing organism.The main work in this paper include as follows:1. The technology of water quality bio-monitoring is presented based on video image understanding techniques. It can continuously obtain video images of testing organism in water quality monitor by computer technology, get behavior parameters of testing organism by tracking the moving objects of video sequence images, according to the analysis of these parameters, achieve the purpose of evaluating water quality.2. Video image understanding techniques are researched. It contains video image preprocessing techniques and video image processing techniques. The video image preprocessing techniques, this paper adopts the typical image filtering techniques, smoothing techniques and enhancement techniques to process the pictures in the process, which is the foundation for the subsequent processing. The video image processing techniques, this paper adopts the typical background reconstruction techniques, object detection techniques and object track techniques to obtain the tracking trajectory of testing organism. In the processing, a mix Gaussian background model based on color features is adopted to reconstructing background, some difference object detection methods are adopted to detecting objects. In this paper, the method based on motion estimation is adopted as moving object track technique. The substance of motion estimation is seeing the moving objects as a particle. This paper realizes objects tracking by the method of using Kalman filter to forecast the particle of moving objects.3. Fish is adopted as the testing organism of water quality bio-monitoring is proposed. Due to fish's some conspicuous advantages, such as low cost, easy access to resource and high sensitivity to water quality etc, made it as our first choice. Fish is used as testing organism in this experiment, software is compiled with C++ program language, simulation experiment environment is built, simple water quality monitoring system is designed. The video monitoring images should be obtained, and then be processed. Ultimately the purpose of evaluating water quality is achieved by obtaining and analyzing the behavior parameters such as average swimming height, average swimming speed of testing organism.