卢焕达¹，刘 鹰²，范良忠¹
（1 浙江大学宁波理工学院 信息科学与工程学院， 宁波 315100；2 中国科学院海洋研究所，青岛 266071）

摘要: 针对传统的鱼类行为观察方法的缺点，提出了一种基于计算机视觉的鱼类行为自动监测系统的设计与实现方法。系统利用摄像设备对鱼的行为进行24小时监控，通过监测鱼类体色与游动速度（加速度）的变化，自动发现由于应激或疾病等引起的行为异常。该文设计的方法对水产养殖业实施水产生物的自动化监测具有一定的借鉴意义。

关键词：计算机视觉；鱼类行为；行为监测

Design and implementation of fish behavior automatic monitoring system
based on computer vision
LU Huanda ¹, LIU Ying ², FAN Liang-zhong ¹
( 1 School of Information Science and Engineering, Ningbo Institute of Technology, Zhejiang University, Ningbo 315100, China;
2 Institute of Oceanology, Chinese Academy of Sciences, Qingdao 266071, China)

Abstract: A design and implementation of fish behavior automatic monitoring system was proposed after analyzing the disadvantages of traditional observation methods. Based on CCD (Charge-coupled Device) camera mounted at the side face of tank, the system was designed to monitor the fish behavior in 24 hours. Abnormal behavior which may be caused by stress state or disease was detected automatically through monitoring the skin change and swimming speed of fish. The method and monitoring system will be helpful for improving production automation in modern aquaculture industry.

Key words: computer vision; fish behaviors; behavior monitor

基金项目：国家自然科学基金项目（30972267）
作者简介：卢焕达（1979—），讲师，主要从事机器学习、机器视觉及数字水产相关研究。E-mail: huandalu@gmail.com
 

(来源：《渔业现代化》2011.38(1)：19-23)