计算机视觉视域中水稻叶片叶绿素含量的数学建模  

The mathematical modeling of chlorophyll content in rice in view of computer vision

作　　者：陈诚[1] 廖桂平[1] 史晓慧[1] 李锦卫[1] 

机构地区：[1]湖南农业大学农业信息研究所,湖南长沙410128

出　　处：《湖南农业大学学报（自然科学版）》2011年第5期474-478,共5页Journal of Hunan Agricultural University(Natural Sciences)

基　　金：国家自然科学基金项目(31071328);湖南省研究生科技创新基金项目(CX2010B280)

摘　　要：利用计算机视觉技术获取水稻叶片的颜色指标DGCI、Hv、I2、I3、(2G-R-B)/L*和Hv*Diff,结合BP网络、多元回归模型和遗传算法,建立叶绿素相对含量(SPAD值)的预测模型,对叶片SPAD值进行数学模拟,模型的数学表达式为Y=purelin[W5*tansig(W4*X,B4),B5]。利用所建立的模型对叶绿素相对含量(SPAD值)进行预测,结果与大田实测数据相对误差率仅为3.355 7%。DGCI, Hv, I2, I3, （2G-R-B）/L＊ and Hv＊Diff, which are the color index of leaves, were acquired by using computer vision technology. Combining those indexes with BP network, multiple regression models and genetic algorithms, the predictive model for chlorophyll relative content was established The value of SPAD for leaves is simulated by mathematics, and the mathematical expressions of model is Y=purelin[Ws＊ tansig（W4＊ X, B4）, B5]. Using the established model to predict the value of SPAD, relative error rate is 3.355 7 % between the results and measured data in field.

关 键 词：计算机视觉 叶绿素相对含量 BP神经网络 多元回归 遗传算法 聚类分析 水稻 

分 类 号：S511[农业科学—作物学] TP183[自动化与计算机技术—控制理论与控制工程]