Implementation of Water Quality Management by Fish School Detection Based on Computer Vision Technology

摘要

To solve the detection of abnormal water quality, this study proposed a biological water abnormity detection method based on computer vision technology combined with Support Vector Machine (SVM). First, computer vision is used to acquire the parameters of fish school motion feature which can reflect the water quality and then these parameters were preprocessed. Next, the sample set is established and the water quality abnormity monitoring model based on computer vision technology combined with SVM is acquired. At last, the model is used to analyze and evaluate the motion characteristic parameters of fish school under unknown water, in order to indirectly monitor the situation of water quality. In view of great influence of kernel function and parameter optimization to the model, this study compared different kinds of kernel function and then made optimization selection using Particle Swarm Optimization (PSO), Genetic Algorithm (GA) and grid search. The results obtained demonstrate that, that method is effective for monitoring water quality abnormity.