文献类型：期刊文献

中文题名：基于红边参数与PCA的GA-BP神经网络估算叶绿素含量模型

英文题名：GA-BP Neural Network Estimation Models of Chlorophyll Content Based on Red Edge Parameters and PCA

作者：李永亮[1] 张怀清[1] 林辉[2]

第一作者：李永亮

机构：[1]中国林业科学研究院资源信息研究所北京100091;[2]中南林业科技大学林业遥感信息工程研究中心长沙410004

年份：2012

卷号：48

期号：9

起止页码：22-29

中文期刊名：林业科学

外文期刊名：Scientia Silvae Sinicae

收录：CSTPCD;;Scopus;北大核心:【北大核心2011】；CSCD:【CSCD2011_2012】；

基金：国家高技术研究发展计划(863计划)(2012AA102001);国家自然科学基金(30871962);高等学校博士学科点专项科研基金(200805380001)

语种：中文

中文关键词：红边参数;GA-BP神经网络;叶绿素含量;模型

外文关键词：red edge parameters; GA-BP neural network; chlorophyll content; models

分类号：S711

摘要：利用便携式ASD野外光谱辐射仪对杉木冠层叶片光谱进行测定,同时以分光光度法对叶片叶绿素含量进行提取。样本经均值处理、平滑处理和微分处理后,进行红边参数提取。对11个红边参数以PCA方法进行降维,将得到的前7个主成分得分作为网络输入参数,叶绿素含量作为网络输出参数,以遗传算法(GA)优化网络初始权值阈值,建立隐含层神经元数分别为4,6,8,10,12和14的6种单隐层BP神经网络模型。以R2,RMSE和相对误差作为模型精度检验标准,结果表明:6种模型预测精度均可达到92.0%以上,其中隐含层神经元数为10时,预测精度最高,可达97.372%。说明此种模型可对杉木冠层叶片叶绿素含量进行高精度估算。
High-precision estimation model of arbor canopy chlorophyll content is important to forestry and ecology. The spectral reflectance of canopy was measured by ASD FieldSpec and the chlorophyll content was measured by spectrophotometry at the same time. The sample data were pretreated by the methods of mean, smoothing and derivative, and then the red edge parameters of samples were extracted from the pretreated spectra data. The eleven red edge parameters were analyzed with principal component analysis （PCA）. The anterior 7 principal components computed by PCA were used as the input variables of back-propagation artificial neural network （BP-ANN） which included one hidden layer which had four, six, eight, ten, twelve or fourteen neurons, while the chlorophyll content was used as the output variables of BP-ANN, and then the three layers BP-ANN discrimination model was built. Weight value and threshold value of this model were optimized by using genetic algorithm. The fitness between the predicted value and the measured value was tested by the determination coefficient, the lowest root mean-square error and the average relative error. The results show that the precisions of six models are all above 92.0% and the precision of the model which had ten hidden layer neurons is 97.372%. The canopy chlorophyll content of Chinese fir can be accurately estimated by using this model.