摘要
选用扫描仪获取水稻叶片的数字图像,运用数字图像处理技术研究不同氮素水平水稻的叶色、纹理和形态特征,尝试建立基于叶色、纹理和形态特征的综合水稻氮素营养诊断模型。研究表明:(1)通过水稻叶绿素a含量和叶片颜色特征之间的相关性分析,得到有效颜色特征参量B,b,b/(r+g),b/r,b/g,同时分析叶绿素a含量随叶位的空间分布特征,选择较为稳定的第三完全展开叶作为指示叶或诊断叶;(2)通过与高光谱遥感数据的相关性分析,从机理上说明基于机器视觉的水稻氮素营养诊断是有理论依据的,是可行的;(3)提取水稻叶片颜色、形态综合特征信息,与YIQ电视信号彩色坐标系统的CB值建立水稻氮营养水平的识别模型,得到4个氮素水平的正确识别率为:N0:74.9%;N1:52%;N2:84.7%;N3:75%。初步研究结果表明基于扫描图像特征的田间水稻氮素水平诊断方法是有优势和前景的,有可能建立一种新的实时、快速、准确的田间水稻氮素营养诊断的方法。
In the present research, the scanner was adopted as the digital image sensor, and a new method to diagnose the status of rice based on image processing technology was established. The main results are as follows： （1） According to the analysis of relations between leaf percentage nitrogen contents and color parameter, the sensitive color parameters were abstracted as B, b, b/（r＋g）, b/r and b/g. The leaf position （vertical spatial variation） effects on leaf chlorophyll contents were investigated, and the third fully expanded leaf was selected as the diagnosis leaf. （2） Field ground data such as ASD were collected simultaneous- ly. Then study on the relationships between scanned leaf color characteristics and hyperspectral was carried out. The results indicated that the diagnosis of nitrogen status based on the scanned color characteristic is able to partly reflect the hyperspectral properties. （3） The leaf color and shape features were intergrated and the model of diagnosing the status of rice was established with calculated at YIQ color system. The distinct accuracy of nitrogen status was as follows： NO： 74. 9%; NI： 52%; N2： 84.7 % ,N3：75 %. The preliminary study showed that the methodology has been proved successful in this study and provides the potential to monitor nitrogen status in a cost--effective and accurate way based on the scanned digital image. Although, some confusion exists, with rapidly increasing resolution of digital platform and development of digital image technology, it will be more convenient for larger farms that can afford to use mechanized systems for site-specific nutrient management. Moreover, deeper theory research and practice experiment should be needed in the future.
出处
《光谱学与光谱分析》
SCIE
EI
CAS
CSCD
北大核心
2009年第8期2171-2175,共5页
Spectroscopy and Spectral Analysis
基金
国家自然科学基金项目(30571112
30800703)
国家"863"高科技研究发展计划项目(2006AA10Z204)
浙江省科技计划项目(2007C2308
2008C33008)资助
关键词
颜色特征
主成分分析
综合特征
Color character
Principal component analysis
Synthesis character
