| 李 铠，常庆瑞，陈 倩，陈晓凯，莫海洋，张耀丹，郑智康.基于连续小波变换耦合CARS算法的冬小麦冠层叶片含水量估算[J].麦类作物学报,2023,(2):251 |
| 基于连续小波变换耦合CARS算法的冬小麦冠层叶片含水量估算 |
| Estimation of Water Content in Canopy Leaf of Winter Wheat Based on Continuous Wavelet Transform Coupled CARS Algorithm |
| |
| DOI：10.7606/j.issn.1009-1041.2023.02.15 |
| 中文关键词: 冬小麦 叶片含水量 高光谱 连续小波变换 竞争适应重加权采样 粒子群算法PSO优化BP神经网络 |
| 英文关键词:Winter wheat Leaf water content Hyperspectral Continuous wavelet transform Competitive adaptation reweighted sampling Particle swarm optimization PSO optimization BP neural network |
| 基金项目:国家863计划项目(2013AA102401-2) |
|
| 摘要点击次数: 1296 |
| 全文下载次数: 848 |
| 中文摘要: |
| 为实现干旱地区冬小麦冠层叶片含水量的快速测定，以陕西省乾县为研究区，基于野外冬小麦冠层高光谱数据和实测叶片含水量，对原始光谱进行连续小波变换（continuous wavelet transform，CWT）后得到的小波能量系数与实测含水量进行相关性分析；并通过竞争性自适应重加权采样（competitive adaptive reweighted sampling，CARS）过滤冗余变量，筛选与叶片含水量相关性较好的波长变量，作为优选变量集；通过粒子群算法(particle swarm optimization,PSO)对BP神经网络模型进行优化，构建冠层叶片含水量预测模型并进行分析。结果表明，从尺度1到尺度10，小波系数与冬小麦叶片含水量整体相关性先升后降，中等分解尺度在光谱波段去除噪声、提高相关性方面最佳；基于CARS优选变量集所建的两种模型中，BP-PSO模型预测能力明显优于普通BP神经网络模型，其决定系数可达0.82，均方根误差为0.86%，相对误差为0.82%。这说明CWT-CARS-BP-PSO耦合算法在提升相关性、过滤冗余波段、提高模型预测精度方面效果显著，可用于冬小麦叶片含水量预测。 |
| 英文摘要: |
| In order to achieve rapid determination of water content in canopy leaves of winter wheat in arid areas,based on field canopy height spectral data and measured canopy leaf water content in Qianxian County,Shaanxi Province,the correlation analysis between the wavelet energy coefficient obtained by continuous wavelet transform and the measured canopy leaf water content was conducted. The redundant variables were filtered by competitive adaptive reweighted sampling (CARS),and the wavelength variables with good correlation with canopy leaf water content were selected as the optimal variable set. BP neural network model was optimized by Particle Swarm Optimization (PSO) to construct and analyze canopy leaf water content prediction model. The results show that the overall correlation between wavelet coefficients and water content from scale 1 to scale 10 increases first and then decreases. The medium decomposition scale is the best in removing noise and improving correlation in spectral band. The modeling results of the two models showed that the prediction ability of BP-PSO model is significantly better than that of the ordinary BP neural network model,with the determination coefficient of 0.82,root mean square error of 0.86%,and relative error of 0.82%. The CWT-CARS-BP-PSO coupling algorithm has a significant effect on improving correlation,filtering redundant bands,and improving the prediction accuracy of the model,which can be used to predict the leaf water content of winter wheat. |
|
查看全文 查看/发表评论 下载PDF阅读器 |
| 关闭 |
基本信息
