摘要
应用可见/近红外光谱技术对茄子叶片进行灰霉病害还未在叶片表面出现病症时的早期检测.采用化学计量学方法建立早期检测模型.主成分分析用于对光谱数据进行降维,得到若干个最重要的主成分.但直接从聚类图中无法进行是否染病的鉴别.因而将其作为变量输入BP神经网络,从而减少了计算量,提高了建模精度.检测结果显示,模型具有良好的检测效果,能够达到100%的识别率,正确率也能达到88%.说明运用可见/近红外光谱技术能够实现当病症还未在叶片表面出现时的快速准确的早期检测,为灰霉病早期检测提供了新的途径.
Visible and near-infrared reflectance spectroscopy （Vis/NIRS） technique was applied in the early detection of grey mold （einerea） on eggplant leaves while the symptom had not appeared. Chemometries was used to build the early detection model. In order to decrease the amount of calculation and improving the accuracy, principle component analysis （PCA） was executed to reduce numerous wavebands into several principle components （PCs） as input variables of BPNNS while the PCs plot of three primary PCs was failed. The performance of the BPNNS model is good with 100% recognition rate and 88% correct rate. Thus, it is concluded that the spectra technology is an available one for the early detection of grey mold on eggplant leaves while the symptom has not appeared and it provides a new method for the early detection of grey mold.
出处
《红外与毫米波学报》
SCIE
EI
CAS
CSCD
北大核心
2007年第4期269-273,共5页
Journal of Infrared and Millimeter Waves
基金
国家自然科学基金(60605011)
国家十一五科技支撑计划项目(2006BAD10A04)
高等学校博士学科点专项科研基金课题(20040335034)
浙江省重大科技攻关(2005C12029)资助项目
关键词
可见/近红外光谱
灰霉病
主成分分析
BP神经网络
Vis/near infrared spectroscopy
grey mold （Cinerea）
principal component analysis （PCA）
BP neural networks （BPNNS）
