资源描述：
《基于种子图像处理的大数目玉米品种形态识别》由会员上传分享，免费在线阅读，更多相关内容在行业资料-天天文库。
1、Seediscussions,stats,andauthorprofilesforthispublicationat:https://www.researchgate.net/publication/250261215DiscriminationofNumerousMaizeCultivarsBasedonSeedImageProcess:DiscriminationofNumerousMaizeCultivarsBasedonSeedImageProcessArticleinACTAAGRONOMICASINICA·
2、November2008DOI:10.3724/SP.J.1006.2008.01069CITATIONSREADS8211author:JinzhongYangQingdaoAgriculturalUniversity10PUBLICATIONS15CITATIONSSEEPROFILESomeoftheauthorsofthispublicationarealsoworkingontheserelatedprojects:Relationsbetweenplantdensityandcropyield,theiri
3、mplicationsinmaizeViewprojectAllcontentfollowingthispagewasuploadedbyJinzhongYangon24April2018.Theuserhasrequestedenhancementofthedownloadedfile.作物学报ACTAAGRONOMICASINICA2008,34(6):1069−1073http://www.chinacrops.org/zwxb/ISSN0496-3490;CODENTSHPA9E-mail:xbzw@china
4、journal.net.cnDOI:10.3724/SP.J.1006.2008.01069基于种子图像处理的大数目玉米品种形态识别1,22222杨锦忠郝建平杜天庆崔福柱桑素平12(青岛农业大学植物科技学院,山东青岛266109;山西农业大学农学院,山西太谷030801)摘要:玉米种子鉴别是种子质量检验和育种实践的重要内容。为了评价通过图像处理采集种子特征进行大数目品种鉴别的可行性,扫描了193个品种各50粒种子图像,建立和检验了由4大类46个种子形态特征及其组合组成的6种识别模型。大小类、形状类、纹理类、颜色类、后3
5、类组合、全部4类组合等模型的品种检出率分别为25%、33%、39%、95%、95%和95%,平均籽粒拒真率分别为90%、90%、86%、45%、47%和42%,认伪率为92%、92%、88%、****46%、48%和43%,且后两个误判率高度正相关(r=0.83~0.91)。机器视觉检测具有成本和速度上的优势,能够用于大数目玉米品种的真伪鉴定,形状+纹理+颜色组合模型最佳,经改进技术识别率可以进一步提高。关键词:玉米;图像处理;品种识别;种子形态;判别分析DiscriminationofNumerousMaizeCul
6、tivarsBasedonSeedImageProcess1,22222YANGJin-Zhong,HAOJian-Ping,DUTian-Qing,CUIFu-Zhu,andSANGSu-Ping12(PlantScience&TechnologyCollege,QingdaoAgriculturalUniversity,Qingdao266109,Shandong;AgronomyCollege,ShanxiAgriculturalUniversity,Taigu030801,Shanxi,China)Abstra
7、ct:Seedidentificationplaysacrucialroleinseedqualitytestingandbreedingprogramsinmaize(ZeamaysL.).Machinevisionofseedsurfacefeaturesperformanceswellbasedonafewexperimentsinmaize.Butthesamplenumbersinthesestudieswereonly3–7cultivars.Tofurtherexaminethefeasibilityof
8、imageprocessapplicationindiscriminatingnumerousmaizeculti-vars,sixmodelswerecreatedandvalidatedbymeansofprinciplecomponentanalysisandstatisticaldiscriminationanalysis
