Fish Behavior Ofthree-dimensional Monitoring Research And Application Based On Computer Vision

the environment,water pollution problems have become more and more serious. This paper puts forward an abiological approach for abnormal water quality monitoring based on three-dimensional behaviour of fish. First of all, by acquiring the fish behavior of three-dimensional motion feature parameters can reflect the water quality status of binocular vision;Then set the sample and obtain the abnormal water quality monitoring model based on BP neural network; Finally, using the model analysis and evaluation of fish behavior characteristic parameters of the unknown water, indirect monitoring of water quality in abnormal condition. The main works are as follows:(1) Build a binocular stereo vision system, to obtain the camera intrinsic and extrinsic parameters. Analysis the common coordinate system and camera model vision, calibration for solving the camera internal and external parameters, lay the foundation for the subsequent three-dimensional reconstruction of moving object trajectories.(2) The method of detection and tracking of the water fish under the water. To simple pretreatment for video sequence images, Then using background difference method used to realize target segmentation and centroid tracking algorithm, the target segmentation is the basis of the trajectory tracking, but tracking is the key to subsequent target motion feature extraction, and track the two-dimensional trajectory of fish.(3) Reconstruction of three-dimensional trajectory. On the basis of the previous chapters, parallax method is used to restore the three-dimensional trajectory of moving object coordinates.(4) The extraction of characteristic parameters of the fish body. Movement characteristic parameter definition of fish body, extraction parameters obtained using trajectory.(5) The water network of computer vision and neural abnormality monitoring based on BP neural network. Using the characteristic parameters of the extraction of fish motion as training samples and test samples, the establishment of neural network of semantic mapping model, the final water quality monitoring. Rapid economic development has also brought the inevitable disaster to...