水稻中稗草光谱分析与识别  

Distinguishing Barnyard-grass from Rice Using Spectrum Analysis Technology

作　　者：陈树人[1] 栗移新[2] 毛罕平[1] 沈宝国[2] 

机构地区：[1]江苏大学江苏省现代农业装备与技术重点实验室 [2]江苏大学生物机电工程研究院

出　　处：《农业机械学报》2008年第9期96-99,共4页Transactions of the Chinese Society for Agricultural Machinery

基　　金：江苏省博士后科研资助计划项目(项目编号:0601014B);江苏大学江苏省现代农业装备与技术重点实验室开放基金资助项目(项目编号:NZ200708)

摘　　要：利用ASD光谱仪在室内分别测量了水稻、稗草在350-2 500 nm波段内的反射率。以各波长点处的反射率与绿色反射峰处（555 nm）的反射率的比值为变量,运用SAS统计软件的STEPDISC过程筛选能够区分作物和杂草的变量;判别模型中加入筛选得到的变量,利用DISCRIM过程进行判别分析。实验结果表明,利用4个波长点比率395/555、535/555、705/555和1105/555可有效地从水稻中识别出稗草,其识别率为100%。红边内波长点705 nm处的反射率与555 nm处反射率的比值对模型贡献最大。The spectral reflectance of rice and barnyard-grass was determined in the range from 350 to 2 500 nm using the Analytical Spectral Device Full Range FieldSpec Pro （ASD） on laboratory. The discrimination analysis was carried out with the statistical software package SAS. The spectral reflectance at the green peak （555 nm） was chosen as denominator, and wavelength ratios were calculated as variables to discriminate. Wavelength ratios were selected using the STEPDISC procedure. With the selected variables, the discrimination models were developed using the DISCRIM procedure in SAS. Four wavelength ratios including 395/555, 535/555, 705/555 and 1 105/555, were t/tilized to gain good classification performance （100% accuracy） for distinguishing barnyardgrass from rice. The ratio of spectral reflectance at 705 nm in the red edge to spectral reflectance at 555 nm contributes more to the discrimination model.

关 键 词：水稻 光谱分析 杂草识别 

分 类 号：O657.3[理学—分析化学] S451[理学—化学]