Fish Monitoring Technology Based On Image Processing

With the development and progress of artificial intelligence, pattern recognition and intelligent man-machine interface technology, intelligent video surveillance has grown into the emerging field of computer vision applications. Intelligent video surveillance is in the situations that without the human intervention, automatic analysis of the camera recording the image sequence by means of computer vision and video analysis method, and to achieve the target location, identification and tracking. Analyzing and judging the behavior of the target in this foundation to complete the daily management and in the event of an exception to react promptly.Traditional fish farming produced backward, inefficient, low-yielding, environmental pollution, industrial fish farming can effectively improve the shortcomings. Industrial fish farming can not develop without the support of high-tech, and in this stage intelligent video surveillance began to comes to underwater monitoring them slowly, making the industrial fish farming efficiency gains, increased yields. This article monitors the swimming speed, swimming acceleration, swimming height of the fish to achieve high production.This paper summarizes the work content by the following three points:(1) By the theoretical analysis and simulation of Canny edge operator, Sobel edge operator and other edge operators, Comprehensive analysis of their advantages and disadvantages, choose the Canny edge operators because of the anti-noise ability. Then based on the original Canny edge operators to improve the algorithm. Gaussian function was used for image smoothing by Canny edge operators, which has strong ability to suppress noise, but will also smooth out some of the high-frequency edge, resulting in edge loss. On this point, thia paper using the Hough transform connection missing edge after Canny edge operator,, and the simulation results is ideal.(2) The track of moving target is tracking each target that by moving target tracking detected primarily, and the target records the tracked fish movement, provide the foudation for the extraction of fish trajectory. This paper tracking fish by Kalman filter tracking, but it will be seen the tracking errors from the simulation results over times, and this paper will make improvements of its limitations. To improve the limitations of the Kalman algorithm used IMM.(3) By tracking the trajectory of the fish to acquire the position data of the fish, and it will be used to reflect the situation fish life by calculating the average velocity, acceleration, altitude of the fish.