Research Of Fish Motion State Monitoring Based On Video

Now, biological water quality monitoring is very important in the water safetyprotection and emergency supervision, and the biological water quality based on fish asthe main subject is increasingly becoming the main development direction of theenvironmental protection. The existing water quality monitoring system based on fish ismainly by real-time monitoring, the analysis of behaviors and motion parameters ofbiological individuals or groups for water quality warning. With the fastest growing of thecomputer vision technology, a fish motion state study based on video is conducted in thepaper.Based on the analysis of fish motion characteristics, the research of the paper is touse the skeleton of the fish body to build a fish’s motion model, use this model to get thefish motion parameters such as velocity, swimming corner, tail dimensionality and so on.Among them, the fast and accurate segmentation of the fish and the accurate extraction ofthe skeleton of fish is the focus of the paper, is the basis of building the fish motion model.The study can be summarized as follows:Based on the studying of the video image segmentation algorithm, the paper proposesa new segmentation algorithm by combining the background subtraction method and thecolor threshold method to get a good segmentation image of fish.Through the introduction of the traditional algorithms of skeleton extraction, bringingforward a rapid method to extract fish’s skeleton based on distance transform. The methodis using to solve the problem that the skeleton position is inaccurate and discontinuity byusing the traditional algorithms of skeleton extraction, and the fish’s skeleton is caughtquickly and accurately.To build a fish’s motion model based on the fish’s skeleton. According to analyzingthe motion features of the fish and on the basis of accurately segmentation of fish’s imageand the accurate extraction of the fish’s skeleton, bringing forward the curve fit by usingthe fish’s skeleton, and building the fish motion model, then analyzing the model to catchthe fish’s motion parameters. By accurately segmenting the image of the fish, rapidly extracting fish’s skeleton,simply building fish’s motion model and analyzing the motion parameters and semanticdesign, the research is to provide the rich data for the biological water quality monitoringsystem.