Research On Fish Movement State Detection System Based On Biological Water Quality Monitoring

In the prevention and control of water pollution,the intelligent and efficient water quality monitoring system plays a crucial role in real-time monitoring and effective governance of water environment.The limitations of physical and chemical monitoring only used for the specific indicators and the continuous sampling are overcome by biological water quality monitoring which can reflect the water quality index directly,quickly and comprehensively.As an important indicator of water quality,fish is more sensitive to the change of water environment parameters,and its motion characteristics directly represent the water quality.The foundation can be laid for the further study of water quality characteristics of fish and the change of water environment by monitoring the real-time variation of the fish motion state parameters,which has an important significance for improving the intelligent level of water quality monitoring.The sub project of biological water quality monitoring is completed in this paper based on industrialization cultivation project of Shaanxi Province Education Department in 2014(the development of intelligent environmental protection system based on water quality remote analysis and scientific decision-making,project number:14JF022).A method of the fish motion detection based on biological water quality monitoring is proposed,a water quality monitoring model is established and a set of fish movement detection system for biological water quality monitoring is developed.The main work of this paper is as follows:1.A method of fish motion target detection in video image is proposed.The background model is quickly built with statistical background modeling algorithm.The background subtraction algorithm is used to separate fish objects and background to get gray images of the fish targets.An adaptive segmentation threshold is obtained by maximum between-class variance algorithm(Otsu),and then image segmentation is used to obtain two value image of fish objects.A filtering algorithm based on morphological filtering and area comparison is proposed to effectively filter the noise and completely preserve the fish targets.The centroid of the target is obtained by centroid detection algorithm of connected regions.2.According to the centroid and area of the fish target regions,a target tracking model based on feature is established to track the fish target in real time.The coordinates of the feature point such as the tail and the head of the fish are obtained.The model is built to calculate the fish movement velocity,acceleration,the frequency of the tail beating,the turning frequency and the frequency of hitting the wall.The water quality detection model is established according to the water quality semantic characteristics of the parameters of the fish movement state.3.According to the above algorithm and model,the software of biological water quality monitoring system is designed based on Lab VIEW.The system introduction,video capture,target detection,record query and water quality evaluation module are completed to achieve the real-time processing for collecting with the camera or the stored videos and alarm when water quality exceeds the standard.4.Taking the red crucian carp as the experimental species and changing pH of the water quality as an example,the experimental scheme is designed to test the performance of the system.4800 frame(20min)fish motion videos are collected under each water quality which includes excessively acid(pH=3.5,4.5,5.5),excessively alkaline(pH=9.5,10.5,11.5)and the optimal range for the growth of red crucian carp(pH=6.5,7,7.5,8),the over standard alarming is achieved when the water quality is excessively acid or alkaline by the system analysis and processing.Taking pH=10.5 as an example,the system alarming about water quality is achieved at 2986th frame(12 minutes and 27 seconds).5.The system error is analyzed,the main error is the loss and increase of the targets in the target detection,the detection accuracy of fish targets in each pH is analyzed,the average accuracy is above 0.91.It is realized the real-time monitoring of water quality change information by detecting the movement state parameters of the fish,reached the requirements of the project and has important reference significance for the further study of fish biological water quality monitoring with the completion of the above work.