首页 | 简介 | 作者 | 编者 | 读者 | Ei(光盘版)收录本刊数据 | 网络预印版 | 点击排行前100篇 | English
杨红珍,张建伟,李湘涛,沈佐锐.基于图像的昆虫远程自动识别系统的研究[J].农业工程学报,2008,24(1):188-192.DOI:
基于图像的昆虫远程自动识别系统的研究
投稿时间:2007-03-19  修订日期:2007-05-17
中文关键词:  昆虫  自动识别  远程识别  图像处理  计算机视觉
基金项目:北京市科技新星资助项目(2004B33),“十一五”科技支撑计划项目(2006BAD08A01)
作者单位
杨红珍 (1972-)山西新绛人副研究员博士主要从事计算机技术在昆虫学方面的应用研究。北京天桥南大街126号 北京自然博物馆100050。Email:hongzhen6072@126.com 
张建伟 中国农业大学农学与生物技术学院北京100094 
李湘涛 北京自然博物馆北京 100050 
沈佐锐 (1944-)教授博士生导师北京海淀区圆明园西路2号 中国农业大学农学与生物技术学院100094。Email: ipmist@ cau.edu.cn 
摘要点击次数: 5817
全文下载次数: 2971
中文摘要:只有对害虫进行鉴定才能在农业生产中对害虫进行有目的的防治,而对昆虫进行鉴定只有少数分类专家才能完成,鉴定需求的日益增加与专家相对较少形成了一对尖锐的矛盾。该文的研究尝试为该矛盾的解决提供一条新的思路:在标准方法下获取昆虫图像,并经由Internet网络上传给自动种类识别系统服务器,从而实现远程识别。系统首先对昆虫图像进行基于形状和颜色特征值的提取。昆虫图像的形态特征值由矩形度、延长度、球状型、叶状型、似圆度和 7个Hu不变矩等12个特征值组成,颜色特征值由红、绿、蓝、灰度真方图及基于红、绿的二维色度直方图特征值分别组成,然后建立径向基神经网络分类器,每一特征向量由独立的径向基神经网络做为分类器,最终识别由每个分类器识别结果的线性组合而成。采用该系统对16种昆虫进行了测试,每种昆虫取40个样本,20个用做训练、20个用做测试,准确率达到96%以上。
Yang Hongzhen,Zhang Jianwei,Li Xiangtao,Shen Zuorui.Remote automatic identification system based on insect image[J].Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE),2008,24(1):188-192.DOI:
Remote automatic identification system based on insect image
Author NameAffiliation
Yang Hongzhen College of Agronomy and Biotechnolog ,China Agricultural University, Beijing 100094, China
Beijing Museum of Natural History, Beijing 100050, China 
Zhang Jianwei College of Agronomy and Biotechnolog ,China Agricultural University, Beijing 100094, China 
Li Xiangtao Beijing Museum of Natural History, Beijing 100050, China 
Shen Zuorui College of Agronomy and Biotechnolog ,China Agricultural University, Beijing 100094, China 
Key words:insect  automatic identification  remote identification  image processing  computer vision
Abstract:Only after the pest was identified, could the purposeful controlling of the pest be conducted in agricultural production. The identification of insects is one of the most important jobs in entomology. A novel method based on computer vision technology with a remote way can make a valuable contribution to reducing the burden of routine identifications. It can also provide a potential solution to the growing burden of routine species identifications presently faced by a dwindling community of expert taxonomists. Insect image was captured and uploaded to identification server from internet. Identification process was done by server automatically and results were given back to user immediately. Shape and color characteristics were used to form the separated pattern vector and RBF network was used as classifier for each pattern. Each classifier was linearly combined to give the final identification result. Shape pattern consists of rectangularity, elongation, roundness, sphericity, lobation and seven movement invariants. R, G, B, L histogram and R.G chromaticity histogram were treated as separated color characteristic vector. The identification system was used to test 16 species, 40 samples for each, with 20 of the 40 samples for “training” and the remainder for “test”. The accuracy reaches above 96%.
查看全文   下载PDF阅读器

京ICP备06025802号-3

主办单位:中国农业工程学会 单位地址:北京朝阳区麦子店街41号

服务热线:010-59197076、59197077 传真: 邮编:100125 Email:tcsae@tcsae.org
本系统由北京勤云科技发展有限公司设计

京公网安备 11010502031390号