Research On Detecting And Tracking Of Fish Group Based On Machine Vision

Biological monitoring techniques for water quality are technology, the basic principle of which is that using the reaction of biological individual, population and group to environmental pollution to expound the pollution state of the environment. The result of monitoring techniques can reflect the change of the water quality of water environment directly and real-time. Using the changes of biodiversity characteristics to monitoring the water quality is an effective way. So, how to get the characteristics of biodiversity behavior is a very important part of the water quality monitoring systemIn order to get the information of characteristics of fish group to make a completed water quality monitoring system, this paper used fish group as the indicator organism of biomonitoring for water quality monitoring, and mainly study the outline of computer vision applications in movement behavior monitoring by fish group. The main content of the article includes: real-time detecting fish group, tracking fish group object and design and realization for the monitoring System and platform. The main results of this paper summarized as follows:First, Study on real-time detection method for the moving objects Based on graph cut. A new moving object detection algorithm based on graph cut. Firstly, a method of watershed transform is adopted to divide the image into parts, and build the net in every part. Then a new energy function was constructed, which contain soft constraints, hard constraints and time constraints. Finally, get the minimum cut by minimizing the energy function by graph cut.Second, Study on fish group tracing Based on delaunay triangulation. In order to accurately track fish group, a new algorithm is proposed, which is based on the traditional background subtraction algorithm and the Delaunay Triangulation network. At first, the traditional background subtraction algorithm is used to deal with the video image sequence; Then, the coordinates of each fish are evaluated; Third, the constraint Delaunay Triangulation is established and some objects are removed so that fish group can be detected. Experimental results show that this algorithm can track fish groups accurately and provide effective data for monitoring water quality.Last, study on the system and platform of fish group detecting and tracking based on computer vision. In order to make the experiment environment to similar to the real environment, the existed experiment platform has been appropriately modified. The architecture of fish group detecting and tracking based on computer vision has been designed and vision processing model and fish group tracking model initially constructed.