Fish activity method based on neural network

Abstract

The invention discloses a fish activity method based on a neural network. The method includes the following steps: 1. real-time monitoring crucian as a biological monitoring target; 2. extracting a target fish outline by using background differencing method, real-time monitoring a crucian population, obtaining a tracking video sequence of the crucian; and 3. automatically obtaining texture characteristics of a to-be-monitored fish which is alive or dead through the neural network, comprising the following steps: 3.1. collecting and extracting the outline information data of a fish target, determining whether the quality of the to-be-inspected water is normal or not based on the collected outline information; 3.2. based on the neural network, training the data and generating a mature classifier structure model; 3.3. using new characteristic data to detect and determine whether the detection model is mature; and 3.4. on-line and real-time detecting water quality data by using the mature detection model, and eventually implementing on-line detection of the toxicity of the water quality.

Description

technical field [0001] The invention relates to the technical field of biological water quality toxicity detection and neural network classification, and proposes a classification method for judging fish activity by using a neural network, which can improve the accuracy of biological water quality toxicity detection. technical background [0002] Normal water quality means that the sensory (transparency, peculiar smell, etc.) and biochemical indicators (phosphorus, nitrogen content, pH, etc.) of water quality are within the allowable range of national standards. Water pollution often changes the sensory and biochemical indicators of water quality. By checking the deviation between the current data indicators of water quality and the normal data indicators, it is possible to determine whether the water quality is abnormal. Therefore, water quality detection is a classification problem, that is, to determine whether the water quality is in a normal or abnormal state according ...

AI Technical Summary

Problems solved by technology

Although the physical and chemical detection methods can accurately give the index data that cause water quality abnormalities, there are still the following shortcomings: 1) Most of the water quality data are collected manually at regular intervals, which leads to data lag and cannot cope with sudden water pollution; 2 ) Due to the use of physical and chemical testing, a series of instruments and equipment for analysis and testing are required, resulting in h

Images