期刊论文

中文

湖南农业大学学报(自然科学版)

1007-1032

2011

37

5

474-478

10.3724/SP.J.1238.2011.00474

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）；

本校为第一机构

利用计算机视觉技术获取水稻叶片的颜色指标DGCI、Hv、12、13、(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[W_5~* tansig(W_4~*X,B_4),B_5]．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3.3557%．

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...