Please wait a minute...
浙江大学学报（农业与生命科学版）
浙江大学学报（农业与生命科学版）  2018, Vol. 44 Issue (4): 490-498    DOI: 10.3785/j.issn.1008-9209.2018.07.030
论文     
基于计算机视觉的岱衢族大黄鱼选育群体外形特征模式识别方法
余心杰1*，吴雄飞2，沈伟良2
（1. 浙江大学宁波理工学院，浙江 宁波 315100；2.宁波市海洋与渔业研究院，浙江 宁波 315010）
Pattern recognition method for the identification of Daiqu large yellow croaker based on computer vision#br#
YU Xinjie1*, WU Xiongfei2, SHEN Weiliang2
(1. Ningbo Institute of Technology, Zhejiang University, Ningbo 315100, Zhejiang, China; 2. Ningbo Marine and Fishery Research Institute, Ningbo 315010, Zhejiang, China)
 全文: PDF(5455 KB)   HTML ( ) 
摘要： 通过计算机视觉测定岱衢族大黄鱼F2、F3 代2 类选育群体的24 个形态参数，利用主成分分析（principalcomponent analysis, PCA）和连续投影算法（successive projections algorithm, SPA）对形态参数进行特征提取和选择，获得PCA变换主元特征、PCA选择特征和SPA选择特征3 组不同的特征变量集，最后以特征变量集为输入建立岱衢族大黄鱼F2、F3代选育群体的稀疏表示识别模型。PCA、SPA特征提取和选择结果表明，全长/体长、全长/头长、全长/尾柄长、体长/头长、尾柄长/尾柄高是反映岱衢族大黄鱼F2、F3代选育群体之间形态差异的主要特征变量。稀疏表示模型的识别结果表明：3 组特征变量集对岱衢族大黄鱼F2、F3代选育群体样本都能较好地进行识别，平均识别准确率为88.3%、79.0%、80.5%；其中PCA变换主元特征对岱衢族大黄鱼F2、F3代的识别准确率最优，为88.3%。本研究结果为建立岱衢族大黄鱼外形指标及开展外形评价提供了有效手段。
关键词： 大黄鱼;  计算机视觉;  群体识别;  形态特征;  稀疏表示    
Abstract: A pattern recognition method based on computer vision was developed for the identification of Daiqu large yellow croaker. First, 24 morphological parameters of Daiqu large yellow croaker were measured for both F2 and F3 generations by computer vision technology. Then principal component analysis (PCA) and successive projections algorithm (SPA) were respectively applied to extract and select the measured morphological parameters, and then obtained three groups of different characteristic variable sets, which were PCA transformed feature, PCA selected feature, SPA selected feature, respectively. Finally, respectively. Finally, sparse representation (SR) models were built for identificating the F2 and F3 generations of Daiqu large yellow croaker by using the extracted morphological features. The results indicated that the main morphological features for the identification of Daiqu large yellow croaker were total length/body length, total length/head length, total length/caudal peduncle length, body length/head length and caudal peduncle length/caudal peduncle height. The results of SR models showed that the three groups of characteristic variable can effectively identify the F2 and F3 generations of Daiqu large yellow croaker, with the average recognition accuracy of 88.3%, 79.0% and 80.5%; among them, the best SR model with PCA transformed feature achieved an average accuracy of 88.3% for the identification of Daiqu large yellow croaker. This study provides an effective way to establish shape index and carry out shape evaluation research of Daiqu large yellow croaker.
Key words: large yellow croaker    computer vision    group recognition    morphological characteristics    sparse representation
出版日期: 2018-09-13
CLC:  S 237  
基金资助: 浙江省宁波市民生科技项目（2013C11026）；浙江省宁波市农业重大科技专项（2017C110002）
通讯作者: 余心杰（https://orcid.org/0000-0002-6462-3931）     E-mail: xjyu1979@163.com
服务　　
把本文推荐给朋友
加入引用管理器
E-mail Alert
RSS
作者相关文章　　
余心杰
吴雄飞
沈伟良

引用本文:

余心杰, 吴雄飞, 沈伟良. 基于计算机视觉的岱衢族大黄鱼选育群体外形特征模式识别方法[J]. 浙江大学学报（农业与生命科学版）, 2018, 44(4): 490-498.

YU Xinjie, WU Xiongfei, SHEN Weiliang. Pattern recognition method for the identification of Daiqu large yellow croaker based on computer vision#br#. Journal of Zhejiang University (Agriculture and Life Sciences), 2018, 44(4): 490-498.

链接本文:

http://www.zjujournals.com/agr/CN/10.3785/j.issn.1008-9209.2018.07.030    或    http://www.zjujournals.com/agr/CN/Y2018/V44/I4/490

[1] . 山西省小麦孢囊线虫的形态学和分子特征分析[J]. 浙江大学学报(农业与生命科学版), 2012, 38(5): 566-574.
[2] 张静, 邓仕槐, 郭航, 沈飞, 古维刚, 李元伟. 聚类分析在神经元形态特征分类中的应用[J]. 浙江大学学报(农业与生命科学版), 2011, 37(5): 493-500.
[3] 赵雷,郭恺,洪文英,郑经武. 浙江省蔬菜上2种根结线虫的形态学鉴定[J]. 浙江大学学报(农业与生命科学版), 2010, 36(6): 623-629.
[4] 章文英　 应义斌. 苹果果梗和表面缺陷的计算机视觉检测方法研究[J]. 浙江大学学报(农业与生命科学版), 2001, 27(5): 583-586.
[5] 蒋淑丽　 王学德. 5个棉花纤维突变体的胚珠形态特征与农艺性状研究[J]. 浙江大学学报(农业与生命科学版), 2001, 27(3): 285-289.