基于机器视觉的鱼类行为特征提取与分析

贾贝贝; 邵振洲; 王瑞; 渠瀛; 张融; 饶凯锋; 姜安; 刘勇; 关永

doi:10.7524/AJE.1673-5897.20161012001

基于机器视觉的鱼类行为特征提取与分析

1. 首都师范大学成像技术北京市高精尖创新中心, 轻型工业机器人与安全验证实验室, 北京 100048;

2. 北京航空航天大学机械工程及自动化学院, 北京 100191;

3. 中国科学院生态环境研究中心环境水质学国家重点实验室, 北京 100085;

4. 田纳西大学电气工程与计算机科学学院, 美国田纳西州 37996;

5 无锡中科水质环境技术有限公司, 无锡 214024

作者简介: 贾贝贝(1991-),男,硕士,研究方向为机器学习,E-mail:2141002083@cnu.edu.cn

基金项目:

863课题(2014AA06A506)；国家自然科学基金青年基金(21307150)；北京市优秀人才培养资助项目(2014000020124G135)；中国科学院科技服务网络计划(STS计划)(KFJ-SW-STS-171)；广东省省级科技计划项目(2016B020240007)

中图分类号: X171.5

Extraction and Analysis of Fish Behavior Based on Machine Vision

1. Beijing Advanced Innovation Center for Imaging Technology, Light Industrial Robot and Security Verification Laboratory, Capital Normal University, Beijing 100048, China;

2. College of Mechanical Engineering and Automation, Beihang University, Beijing 100048, China;

3. State Key Laboratory of Environmental Aquatic Chemistry, Research Center for Eco-environmental Sciences, Chinese Academy of Sciences, Beijing 100085, China;

4. Department of Electrical Engineering and Computer Science, The University of Tennessee, Tennessee 37996, USA;

5 Wuxi CASA Environmental Technology Co., Ltd, Wuxi 214024, China

Fund Project:

摘要: 近年来，水污染问题备受关注。生物式水质监测成为目前国家环境保护工作的重要任务之一。为准确监测水质污染情况，本文以青鳉鱼(Oryzias latipes)为研究对象，采用非接触式的机器视觉监测技术，提取青鳉鱼的生理特征(呼吸频率)和运动特征(胸鳍和尾鳍的摆动频率)，并分析这些特征与水质之间的关系。本文采用支持向量机(Support Vector Machine, SVM)准确提取鱼鳃，并根据鱼鳃呼吸面积大小变化计算出鱼的呼吸频率。基于形态学细化算法提取青鳉鱼骨架，求出胸鳍和尾鳍的摆动频率。结果显示：不同浓度铜离子暴露实验测得的青鳉鱼生理特征和运动特征与实际情况一致；通过对不同铜离子浓度下的毒性实验数据对比，发现了青鳉鱼的生理特征和运动特征会随不同的铜离子浓度发生相应变化，可以作为水质监测的评价标准。

Abstract: Recently, the problem of water pollution is drawing the increasing attention. The research of biological water quality monitoring has become one of the important tasks to protect the national environment. In order to accurately monitor the water pollution, the medaka is chosen as the research object in this paper. The non-contact machine vision monitoring technology is employed to extract the physiological characteristics (respiratory frequency) and movement characteristics (swing frequency of pectoral fins and tail fin), and analyze the relationship between these characteristics and water quality. The respiratory frequency is firstly calculated based on the area change of gills, which are classified using SVM. Secondly, the medaka skeleton is extracted based on morphological thinning algorithm to find the swing frequency of pectoral fins and tail fin. The results show the exposure experiments are performed using the different concentrations of copper ion to measure the physiological characteristics and movement characteristics. The experimental results are consistent with the actual ones. Meanwhile, when the concentration of copper ion changes, the physiological characteristics and movement characteristics of medaka change correspondingly. It is indicated that the behaviors of medaka can be used as the evaluation criteria for the water quality monitoring.

Key words:

基于机器视觉的鱼类行为特征提取与分析

作者简介: 贾贝贝(1991-),男,硕士,研究方向为机器学习,E-mail:2141002083@cnu.edu.cn
1. 首都师范大学成像技术北京市高精尖创新中心, 轻型工业机器人与安全验证实验室, 北京 100048;

2. 北京航空航天大学机械工程及自动化学院, 北京 100191;

3. 中国科学院生态环境研究中心环境水质学国家重点实验室, 北京 100085;

4. 田纳西大学电气工程与计算机科学学院, 美国田纳西州 37996;

5 无锡中科水质环境技术有限公司, 无锡 214024

收稿日期: 2016-10-12

基金项目:

关键词:

Extraction and Analysis of Fish Behavior Based on Machine Vision

1. Beijing Advanced Innovation Center for Imaging Technology, Light Industrial Robot and Security Verification Laboratory, Capital Normal University, Beijing 100048, China;

2. College of Mechanical Engineering and Automation, Beihang University, Beijing 100048, China;

3. State Key Laboratory of Environmental Aquatic Chemistry, Research Center for Eco-environmental Sciences, Chinese Academy of Sciences, Beijing 100085, China;

4. Department of Electrical Engineering and Computer Science, The University of Tennessee, Tennessee 37996, USA;

5 Wuxi CASA Environmental Technology Co., Ltd, Wuxi 214024, China

Received Date: 2016-10-12

Fund Project:

Keywords:

全文HTML

参考文献 (0)