| Summary: | 碩士 === 佛光大學 === 資訊學系 === 99 === Industrial property developed, resulting in environmental load is increasing, relatively more vulnerable to water pollution, and water companies to ensure the public safety of water, in the setting of the water quality of raw water intake station fish toxicity monitoring system. This system will use the biological characteristics of long-term survival in the water, when water quality changes produced, the water will produce abnormal biological behavior even fatal reaction, so often used in aquatic toxicity of the water quality monitoring system, and because many aquatic species this paper will use the zebra fish, two kinds of fish species and Zhu Wenjin toxicity water
quality monitoring.
Water company's traditional set of fish toxicity of the raw water quality monitoring systems approach is a site monitoring by the staff of the behavioral response of fish for 24 hours of continuous monitoring, however, due to human factors, so the monitors can not be 24 hours of continuous monitoring of fish behavior reaction, so in order to allow water quality monitoring system to achieve the purpose of automated water quality monitoring system, the use of image processing technology to solve this problem. However, image processing technology for automatic identification of fish posture will become an important research topic.
In this paper the use of water exposed to contaminated fish produce behavioral responses, make the appropriate warning indicators. System architecture consists of three main modules: the prospects to detect, identify fish and fish behavior analysis profile. Enter the real-time video, using W4 method to establish the background model, and then captured by the background subtraction and calculate the fish image and the center of its fish and related information, will capture the fish of the images selected image feature points, and images from the selected feature points after the use of support vector machine to establish the direction of the fish identification model, and then to fish gesture recognition. Finally, fish behavior analysis, namely the use of fish of the abnormal behavior of the reaction, through the fuzzy theory to calculate the current level of water contamination of the situation. The results show that the use of image processing paper, support vector machines with fuzzy theory and technology to achieve automation of the raw water toxicity monitoring system purposes, and also to solve the traditional use of manual monitoring behavioral
responses of fish problems.
|
|---|
