基于PROSPECT模型的蔬菜叶片叶绿素含量和SPAD值反演

引用本文

雷祥祥, 赵静, 刘厚诚, 张继业, 梁文跃, 田佳灵, 龙拥兵. 基于PROSPECT模型的蔬菜叶片叶绿素含量和SPAD值反演[J]. 光谱学与光谱分析, 2019,39(10): 3256-3260.
LEI Xiang-xiang, ZHAO Jing, LIU Hou-cheng, ZHANG Ji-ye, LIANG Wen-yue, TIAN Jia-ling, LONG Yong-bing. Inversion of Chlorophyll Content and SPAD Value of Vegetable Leaves Based on PROSPECT Model[J]. Spectroscopy and Spectral Analysis, 2019,39(10): 3256-3260.
Doi:10.3964/j.issn.1000-0593(2019)10-3256-05 复制到剪切板

Permissions

光谱学与光谱分析所有

基于PROSPECT模型的蔬菜叶片叶绿素含量和SPAD值反演

雷祥祥¹, 赵静¹, 刘厚诚², 张继业², 梁文跃¹, 田佳灵², 龙拥兵^1,^*

1. 华南农业大学电子工程学院, 广东广州 510642

2. 华南农业大学园艺学院, 广东广州 510642

*通讯联系人 e-mail: yongbinglong@126.com

作者简介: 雷祥祥, 1994年生, 华南农业大学电子工程学院研究生 e-mail: ilei00@foxmail.com

收稿日期: 2018-08-16 接受日期: 2018-12-15

基金: 广东省重点领域研发计划项目(2019B020219002, 2019B020214005), 广东省2019年省级乡村战略振兴专项(粤财农(2019)73号), 国家自然科学基金项目(11774099)资助

摘要

叶绿素含量是衡量植物营养和病虫害发生情况的重要指标。传统的分光光度法对植物叶片破坏性较大且无法实时、快速、无损地获取叶绿素含量。新兴的利用叶绿素仪测量叶绿素相对含量(以下简称SPAD值)的方法不能定量获取实际含量。光学辐射传输模型PROSPECT从生物物理、化学的角度以及能量传输的过程出发, 定量描述了叶片色素、水分、结构参数等对叶片反射光谱的影响。因此, 提出利用PROSPECT模型同时反演蔬菜叶片叶绿素含量和SPAD值, 实时、快速、无损、定量获取植物叶片叶绿素的含量。第一, 多次测量三种蔬菜叶片的反射光谱, 并用叶绿素仪测量SPAD值。然后, 预处理光谱数据, 获得平均反射率光谱。第二, 以欧式距离为评价函数, 利用PROSPECT模型对实测反射率光谱进行拟合。拟合过程中三种蔬菜欧式距离最大为0.008 9, 最小为0.006 4, 平均为0.007 5, 表明该模型能够很好地拟合蔬菜叶片的反射率光谱。第三, 根据拟合结果, 反演叶绿素含量和透射率光谱, 再根据透射率光谱获取叶片在940和650 nm波长处的光透过率, 计算叶片的反演SPAD值。第四, 建立反演叶绿素含量、反演SPAD值与实测SPAD值的关系模型。结果表明: (1)利用该模型反演得到的叶绿素含量值与实测SPAD值有较好的线性关系, 其关系模型为: y=1.463 3 x+16.374 3, 两者相关系数为0.927 1, 模型的决定系数为0.862, 均方根误差为2.11; (2)利用该模型反演得到的SPAD值与实测SPAD值之间线性关系较好, 其关系模型为: y=0.986 9 x-0.668 3, 两者相关系数为0.845 1, 模型的决定系数为0.714 3, 均方根误差为3.380 2。研究表明, 通过测量植物叶片的反射率光谱, 利用PROSPECT模型可以无损、定量地获取蔬菜叶片的叶绿素含量和SPAD值。该方法可推广至其他植物的叶绿素测量和实时监测, 为变量施肥、精准种植提供可靠的数据支持。研究结果对蔬菜生长态势的无损监测具有重要的意义。

关键词: 高光谱; 蔬菜叶片; 叶绿素含量; SPAD值; 反演

中图分类号:O657.3 文献标志码:A

Inversion of Chlorophyll Content and SPAD Value of Vegetable Leaves Based on PROSPECT Model

LEI Xiang-xiang¹, ZHAO Jing¹, LIU Hou-cheng², ZHANG Ji-ye², LIANG Wen-yue¹, TIAN Jia-ling², LONG Yong-bing^1,^*

1. School of Electronic Engineering, South China Agricultural University, Guangzhou 510642, China

2. School of Horticulture, South China Agricultural University, Guangzhou 510642, China

*Corresponding author

Abstract

Chlorophyll content is an important indicator for evaluation of plant nutrition and the occurrence of the pests. The traditional spectrophotometry causes damage to plant leaves and can not be used to obtain the chlorophyll content in a real time, fast and non-destructive way. The chlorophyll meter is recently developed to measure the relative content of chlorophyll (referred to as SPAD value). But this method cannot be used to quantitatively obtain the actual content. The optical radiation transmission model PROSPECT can quantitatively describe the effects of leaf pigment, moisture and structural parameters on the reflection spectrum of the leaves with careful consideration of the biophysics, chemistry and the process of energy transfer in the leaves. In the paper, therefore, this model is used to simultaneously inverse the chlorophyll content and SPAD value of vegetable leaves, and obtain the chlorophyll content of plant leaves real-timely, quickly, non-destructively and quantitatively. First, the reflection spectra of the leaves for three vegetables were measured several times, and the SPAD values of these leaves were measured with a chlorophyll meter. Then, the spectral data were preprocessed to obtain the average reflectance spectrum. Second, the averaged reflectance spectra were fitted by the PROSPECT model with the Euclidean distance as the evaluation function. The maximum distance of the three vegetables in the fitting process was 0.008 9, the minimum was 0.006 4, and the average was 0.007 5. Such a low Euclidean distance indicated that the model could well fit the reflectance spectrum of vegetable leaves. Thirdly, according to the fitting results, the chlorophyll content and the transmittance spectrum were inversed, and the inversed SPAD value of the leaves was calculated with the light transmittance of the leaves at 940 and 650 nm as input parameters. Fourth, this paper established a relationship model between the inversed chlorophyll content, the inversed SPAD value and the measured SPAD value. Two main results were obtained: (1) the chlorophyll content obtained by the model has a good linear relationship with the measured SPAD value and the relationship model is y=1.463 3 x+16.374 3. The correlation coefficient between them is 0.927 1. The coefficient of determination of the model is 0.862 and the root mean square error is 2.11. (2) A good linear relationship of y=0.986 9 x-0.668 3 is obtained for the inversed SPAD value and the measured SPAD value. The correlation coefficient between them is 0.845 1.The coefficient of determination of the model is 0.714 3 and the root mean square error is 3.380 2. The research shows that the PROSPECT model can be used to obtain the chlorophyll content and SPAD value of vegetable leaves nondestructively and quantitatively with the measured reflectance spectrum of plant leaves as input parameters. This method can be extended to other plants for chlorophyll measurement and real-time monitoring and can provide reliable data support for variable rate fertilization and precision planting. The results presented in this paper can be applied to monitor the growth of the vegetables nondestructively.

Keyword: Hyperspectral; Vegetable leaf; Chlorophyll content; SPAD value; Inversion

文章图片

引言

绿色蔬菜富含蛋白质、碳水化合物、维生素以及各种微量元素, 能够对人类健康产生较大影响。其中, 叶绿素含量是植物光合作用能力和营养信息表达的一个重要指标, 广泛用于监测蔬菜的健康状况^{[1, 2]}。传统的植物叶片叶绿素含量检测方法主要为分光光度法。该方法应用较为广泛, 被科研工作者用于测定各类植物叶片中叶绿素a、叶绿素b的含量。例如, Cui等^[3]利用改进的分光光度法测定水体中叶绿素a, 取得了比传统方法更准确的测量结果。但该方法为化学方法, 会对植物叶片产生破坏, 且检测过程还含有丙酮等有害试剂。

SPAD叶绿素相对浓度测量仪利用叶片在940和650 nm波长处的光透过率之比来计算叶绿素相对含量, 可以快速检测叶片叶绿素相对含量。例如, Hu等^[4]利用SPAD仪确定了黄瓜氮素诊断的最佳叶位, 并表明SPAD值和叶片含氮量之间存在显著的相关性。但该方法不足之处在于不能定量地检测叶绿素含量。

近年来, 高光谱技术的发展为检测植物的健康状况提供了新的方向。 Ding等^[5]提取7个能反映叶绿素含量变化的最优参量, 建立了温室番茄叶片叶绿素含量预测模型, 该模型具备较好的预测能力。 Li等^[6]利用红边区域特性, 提出了一种新的提取反射光谱的方法, 能很好的估算水稻和小麦叶片的叶绿素含量。 Lu等^[7]利用生长季节樱桃叶的多种光谱变量, 评价了高光谱遥感在估算叶绿素a、 b和叶绿素a+b浓度的能力, 实验结果的均方根误差较小。然而, 这些模型易受到植物叶片结构和观测角度等因素的影响。 Baret等^[8]建立光学辐射传输模型PROSPECT, 该模型定量描述了叶片色素、水分、结构参数等因素对叶片反射光谱的影响, 可以定量获取叶片内部各种成分的含量。例如, Jiang等^[9]比较植被指数与四个版本的PROSPECT模型反演性能, 发现两者性能类似, 但PROSPECT能够提供表面反射率和叶片内部结构等附加特性。 Sun等^[10]通过PROSPECT模型研究了采用32个通道的HSL系统估算叶片叶绿素含量的可能性, 结果表明PROSPECT模型适用于叶绿素含量的反演。目前, 利用光学辐射传输模型PROSPECT模型同时反演植物叶片叶绿素含量、 SPAD值的相关研究则比较少。本研究将利用该模型拟合蔬菜叶片的反射率光谱, 进而反演叶片的叶绿素含量和SPAD值, 在此基础上建立反演叶绿素含量、 SPAD值与实测SPAD值的关系模型。

1 实验部分

1.1 光谱数据采集

本研究以华南农业大学设施园艺实验基地栽培的绿色奶油生菜、罗马生菜、广府35菜心为研究对象。在实验基地采摘奶油生菜、罗马生菜、广府35菜心后, 立即送往实验室测量光谱数据。在光谱数据测量过程中, 首先利用日本MINOLTA公司的SPAD-502叶绿素检测仪测量蔬菜叶片的SPAD值, 然后摘取叶片, 利用光纤光谱仪(Flame-S-VIS-NIR-ES型, 海洋光学)测量叶片同一位置的反射光谱。具体测量方法如下, 首先扫描无光环境下仪器噪声光谱B₁, 然后打开光源, 扫描白色标准板的反射光R₀; 然后扫描植物叶片的反射光R₁, 测量叶片的背景噪声B₂, 植物叶片的反射率光谱可表示为^[11]

$R = \frac{R_{1} - B_{1}}{R_{0} - B_{2}}$ (1)

1.2 光谱模拟

1.2.1 PROSPECT模型

本研究采用PROSPECT-D拟合植物的反射率光谱。该模型预测不确定性较小, 能实现对植物叶片叶绿素、胡萝卜素、花青素等的高分辨率监测; 实验表明, 该模型版本优于之前版本, 能在可见光波段以最小的误差模拟真实叶片的光学性质, 在近红外波段、短波红外波段的误差和其他版本类似^[12]。该模型定量描述了植物叶片反射率光谱、透过率光谱与模型参数N(结构参数)、 C_ab(叶绿素含量)、 C_car(胡萝卜素含量)、 C_anth(花青素含量)、 C_brown(棕色物质含量)、 C_w(等效水含量)、 C_m(干物质含量)之间的关系。

1.2.2 PROSPECT模型敏感度分析

敏感度可以描述某一变量以相同步长变化的过程中对模型输出结果的影响程度。以X₀为参考点, 定义敏感度计算公式为^[13]

$F' = \overset{n}{\sum_{pert = 1}} \frac{(ρ_{0_{j}} - ρ_{per t_{j}})}{ρ_{0_{j}}}$ (2)

式中 $ρ_{0_{j}}$ 和 $ρ_{per t_{j}}$ 分别为初始参考点X₀和X₀+Δ _X处波长j对应的叶片的反射率。计算PROSPECT模型每个参数的敏感度时, 相对于X₀, 取适当的Δ _X为步长, 其他参数保持不变。计算不同参数的灵敏度时, Δ _X均以相等的相对间隔取值^[14]。本论文中, Δ _X采用初始参考点X₀的0.1倍进行取值。通过PROSPECT模型参数的敏感性计算可以分析各种生化参数对叶片光谱的影响程度, 同时结合模型参数参考范围确定光谱模拟时各生化参数的范围。

1.2.3 叶片叶绿素含量和SPAD值反演

为了反演叶片的叶绿素浓度和SPAD值, 本研究采用Nelder-Mead单纯形直接搜索算法来拟合反射率光谱。该算法是一种局部搜索算法, 在每次迭代过程中, 求出一个较好的探索点, 除去最差的点, 并用探索点来代替, 形成新的单纯形^[15]。通过该算法对模型输入参数的不断调整优化, 可以获得最优输入参数。拟合过程中, 采用模拟值与实测值的欧式距离作为评价函数。欧式距离是常用的距离计算方式, 衡量多维空间中两点之间的绝对距离。其反映了空间中样本之间的不相似性^[16], 适用于计算本研究中光谱拟合的准确性。

$D = \overset{N}{\sum_{i = 1}} \sqrt[]{(r_{1 i} - r_{2 i})^{2}}$ (3)

其中r₁_i为在当前参数矢量下PROSPECT模型模拟的光谱反射率向量, r₂_i为实测光谱反射率。

根据最优拟合结果, 由最佳输入参数获得叶片的叶绿素含量, 由模型输出获得叶片的透过率光谱。根据透过率光谱, 利用下式计算植物叶片的SPAD值^[17]

$SPAD = K \times \lg (T_{1} / T_{2}) + C$ (4)

其中K和C为常数, T₁为叶片在940 nm透过率, T₂为叶片在650 nm处的透过率。

2 结果与讨论

2.1 光谱模拟结果

2.1.1 光谱反射率预处理

本文测量了绿色奶油生菜、罗马生菜、广府35菜心三种蔬菜的反射率。每种蔬菜随机的选择10个叶片, 每片叶取12个采样点测光谱, 因此共360个光谱数据。为降低测量的随机误差, 需要对同一叶片测得的不同点的光谱数据进行光谱平均。首先使用式(1)对测得的每种叶片光谱进行预处理, 并计算同一叶片的平均光谱。预处理后的蔬菜平均反射率光谱如图1所示。

	Figure Option View Download New Window
	图1 蔬菜叶片平均光谱反射率Fig.1 Average spectral reflectance of vegetables

2.2 模型敏感度分析与光谱拟合结果

敏感度分析、光谱模拟的相关参数如表1所示, 计算结果如图2所示。从该图可以看出: 可见光波段(400~780 nm), 叶绿素含量、结构参数、胡萝卜素含量影响较大; 近红外短波波段(780~1 100 nm), 结构参数、干物质含量影响较大; 近红外长波波段(1 100~2 500 nm), 水含量、结构参数、干物质含量影响较大。根据敏感度分析可知, C_w, C_m, C_anth, C_brown对400~780 nm波段的反射率光谱影响很小。根据敏感度分析结果, 结合PROSPECT模型自带的参数范围以及数据拟合情况, 设置光谱模拟的主要参数范围, 如表1所示。

表1 敏感度分析相关参数范围及光谱拟合参数 Table 1 Relevant parameters in the sensitivity analysis and parameters for reflection spectrum simulation of the leaves

确定光谱拟合参数区间和范围后, 利用PROSPECT模型拟合植物叶片反射率光谱, 光谱拟合时对应的欧式距离最大为0.008 9, 最小为0.006 4, 平均为0.007 5, 这表明光谱拟合度较高。部分蔬菜模拟光谱如图3所示。

	Figure Option View Download New Window
	图2 PROSPECT-D模型相关参数敏感性分析 (a): 敏感度分析曲线; (b): 敏感度分析柱状图Fig.2 Sensitivity analysis of the relative parameters in the PROSPECT-D model (a): Sensitivity analysis curve; (b): Sensitivity analysis histogram

	Figure Option View Download New Window
	图3 蔬菜叶片平均光谱拟合 (a): 绿色奶油蔬菜; (b): 罗马生菜平均光谱的模拟光谱; (c): 广府35菜心Fig.3 Simulated spectrum of average spectrum of vegetable leaves (a): Green cream vegetable; (b): Roman lettuce; (c): Guangfu 35 vegetable

2.3 叶绿素含量及SPAD值模拟结果

本文使用PROSPECT模型结合Nelder-Mead单纯形直接搜索算法拟合蔬菜叶片的平均反射率光谱。根据最优拟合结果, 由最佳输入参数获得叶片的叶绿素含量, 并由PROSPECT模型输出获得叶片的透过率光谱。拟合过程中, 不同植物或不同营养情况阶段的结构参数N差异较大。其中, 绿色奶油蔬菜结构参数平均值为2.0, 罗马蔬菜结构参数平均值为1.7, 绿色广府35菜心结构参数平均值为1.4。根据式(4), 利用透过率光谱即可计算植物叶片的SPAD值。

本文研究了反演叶绿素浓度与实测SPAD值的关系。如图4所示, 反演叶绿素浓度与实测SPAD值之间存在良好的线性关系, 其关系模型为: y=1.463 3x+16.374 3。反演叶绿素含量值与实测SPAD值之间的相关系数为0.927 1; 模型的决定系数为0.862, 均方根误差为2.11。

进一步研究了蔬菜叶片反演SPAD值与实测SPAD值的线性关系, 得到的关系模型为: y=0.986 9x-0.668 3。模拟SPAD值与实测SPAD值的相关系数为0.845 1; 模型的决定系数为0.714 3, 均方根误差为3.380 2。从图5可以看出, 反演SPAD值与实测SPAD值之间的线性关系较好。

	Figure Option View Download New Window
	图4 反演叶绿含量与实测SPAD值关系Fig.4 Relationship betweenthe inversed C_ab and measured SPAD

	Figure Option View Download New Window
	图5 反演SPAD值与实测SPAD的关系Fig.5 Relationship between the inversed SPAD and the measured SPAD

3 结论

利用光学辐射传输模型PROSPECT模型, 以欧式距离作为评价函数, 拟合蔬菜叶片反射率光谱, 反演叶绿素含量和SPAD值。根据拟合结果, 利用该模型得到的反演叶绿素含量、反演SPAD值分别与实测SPAD值具有较好的线性关系。 (1)前者关系模型为: y=1.463 3x+16.374 3, 相关系数为0.927 1, 模型的决定系数为0.862, 均方根误差为2.11; (2)反演SPAD值与实测SPAD值关系模型为: y=0.986 9x-0.668 3, 两者相关系数为0.845 1, 模型的决定系数为0.714 3, 均方根误差为3.380 2。以上结论表明, 本文所提出的基于PROSPECT模型的蔬菜叶片叶绿素含量和SPAD值反演方法, 具有实时、快速、无损、定量的特点, 可推广至其他植物的叶绿素测量和实时监测, 为变量施肥、精准种植提供可靠的数据支持。

参考文献

文献列表

[1]	Wang Heng, Qian Xiangjie, Zhang Lan, et al. Frontiers in Plant Science, 2018, 9. [本文引用:1]
[2]	Dimitre A Ivanov, Mark A Bernards. Planta, 2016, 243(1): 263. [本文引用:1]
[3]	CUI Jian-sheng, LÜ Peng-yi(崔建升, 吕鹏翼). Physical Testing and Chemical Analysis Part B: Chemical Analysis(理化检验·化学分册), 2015, 51(1): 105. [本文引用:1]
[4]	Hu Jing, Li Chenxiao, Wen Yifang, et al. IOP Conference Series: Earth and Environmental Science, 2018, 108(2). [本文引用:1]
[5]	DING Yong-jun, LI Min-zan, AN Deng-kui, et al(丁永军, 李民赞, 安登奎, 等). Transactions of the Chinese Society of Agricultural Engineering(农业工程学报), 2011, 27(5): 244. [本文引用:1]
[6]	Li Dong, Cheng Tao, Kai Zhou, et al. ISPRS Journal of Photogrammetry and Remote Sensing, 2017, 129. [本文引用:1]
[7]	Lu Xia, Peng Hongchun. Journal of the Indian Society of Remote Sensing, 2015, 43(1): 109. [本文引用:1]
[8]	Frédéric Baret, Vern C Vand erbilt, Michael D Steven, et al. Remote Sensing of Environment, 1994, 48: 253. [本文引用:1]
[9]	Jiang Jingyi, Comar Alexis, Burger Philippe, et al. Plant Methods, 2018, 14. [本文引用:1]
[10]	Sun Jia, Shi Shuo, Yang Jian, et al. Remote Sensing of Environment, 2018, 212. [本文引用:1]
[11]	SUN Jun, JING Xia-ming, MAO Han-ping, et al(孙俊, 金夏明, 毛罕平, 等). Transactions of the Chinese Society of Agricultural Engineering(农业工程报), 2014, 30(10): 167. [本文引用:1]
[12]	Féret J-B, Gitelson A A, Noble S D, et al. Remote Sensing of Environment, 2017, 193. [本文引用:1]
[13]	Gu Chengyan, Du Huaqiang, Mao Fangjie, etal. International Journal of Remote Sensing, 2016, 37(22): 5270. [本文引用:1]
[14]	Asim Banskota, Shawn P Serbin, Rand olph H Wynne, et al. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2015, 8(6): 3147. [本文引用:1]
[15]	SU Hong-sheng, YIN Kai-le(苏宏升, 殷凯乐). Computer Engineering and Applications(计算机工程与应用), 2016, 52(24): 50. [本文引用:1]
[16]	YANG Ren-jie, YANG Yan-rong, DONG Gui-mei, et al(杨仁杰, 杨延荣, 董桂梅, 等). Spectroscopy and Spectral Analysis(光谱学与光谱分析), 2014, 34(8): 2098. [本文引用:1]
[17]	Casa R, et al. The Journal of Agricultural Science, 2015, 153(5): 876. [本文引用:1]
[18]	Angela De Santis, Emilio Chuvieco, Patrick J Vaughan. Remote Sensing of Environment, 2009, 113(1): 126. [本文引用:1]
[19]	GU Cheng-yan, DU Hua-qiang, ZHOU Guo-mo, et al(谷成燕, 杜华强, 周国模, 等). Chinese Journal of Applied Ecology, 2013, 24(8): 2248. [本文引用:1]

2018

0.0

... 其中, 叶绿素含量是植物光合作用能力和营养信息表达的一个重要指标, 广泛用于监测蔬菜的健康状况^[1,2] ...

2016

0.0

... 其中, 叶绿素含量是植物光合作用能力和营养信息表达的一个重要指标, 广泛用于监测蔬菜的健康状况^[1,2] ...

2015

0.0

... 例如, Cui等^[3]利用改进的分光光度法测定水体中叶绿素a, 取得了比传统方法更准确的测量结果 ...

2018

0.0

... 例如, Hu等^[4]利用SPAD仪确定了黄瓜氮素诊断的最佳叶位, 并表明SPAD值和叶片含氮量之间存在显著的相关性 ...

2011

0.0

DING

Yong-jun

, LI

Min-zan

, AN

Deng-kui

, et al(丁永军, 李民赞, 安登奎, 等). Transactions of the Chinese Society of Agricultural Engineering(农业工程学报), 2011, 27(5): 244.

In order to estimate chlorophyll content of greenhouse tomato leaves fast and accurately and improve the precision management of tomato crop, this research studied the correlation of the chlorophyll content and spectral response at different growth stages of greenhouse tomato. Leaf spectral measurements from each treatment (4 N levels: 0%, 33.3%, 66.6%, 100%) were taken in the greenhouse using ASD FieldSpec HH. Chlorophyll content of tomato leaves were measured by alcoholic-acetone extraction in chemistry lab. It was found that Chlorophyll content of tomato leaf was increasing continuously to the maximum 50 days after the transplantation, while red edge moved to direction of NIR(long wave), green peak position moved to direction of blue light(short wave) and green peak amplitude decreased. The chlorophyll content would decrease after fruiting stage, while red edge, green peak position and its amplitude moved to the opposite direction. For quantitative analysis the relationship between chlorophyll content and spectral response, red edge parameters (Sred(area of red edge), Dred (amplitude of red edge) and Pred (position of red edge ) ) in the first derivative reflectance curve were obtained at bands of 680 nm to 760 nm. Similarly, blue edge, green peak and red valley parameters were defined to reflect spectral character. Vegetation indices have been used extensively to estimate the vegetation growth status. So the following wavelength were used for developing RVI,NDVI and ARVI index: l440 nm, l500 nm, l550 nm, l680 nm, l770 nm, Pblue (position of blue edge), Pyellow (position of yellow edge), Pred (position of red edge), Pgreenpeak (position of green peak), Predvalley (position of red valley). Seven optimal spectral characteristics parameters were chosen with the method of Karhunen-Loeve from the above-mentioned 68 self-defined feature parameters. At last, stepwise multiple regression (SMLR), principal component regression (PCR), ridge regression (RR) and partial least squares regression (PLSR) were used to develop the prediction models of the chlorophyll content of tomato leaf. The best model was obtained by RR. Root MSE was 0.406 and R-Square was 0.839.

为了快速、准确估测温室番茄叶片叶绿素含量，提升作物精细管理水平，利用光谱分析技术研究了温室番茄不同生长阶段叶绿素含量和响应光谱的相关性，在幼苗营养生长阶段叶片叶绿素含量呈增长趋势，到移植50天前后达到最大值，在此期间反射光谱的红边会向红外方向（长波）偏移，同时绿峰向蓝光（短波）方向偏移，绿峰幅值减小。从结果期开始叶绿素含量呈下降趋势，而红边、绿峰及绿峰幅值向相反方向变化。为了定量分析叶绿素含量和叶片反射光谱间的关系，从自定义的68个光谱特征参数中提取了7个能反映叶绿素含量变化的最优参量，并使用逐步回归、岭回归、主成分分量回归和偏最小二乘回归消除了最优参量的多重共线性，建立了叶绿素含量预测模型，其中岭回归模型精度最佳，均方根误差（RMSE）为0.406，决定系数(R2)为0.839。

... Ding等^[5]提取7个能反映叶绿素含量变化的最优参量, 建立了温室番茄叶片叶绿素含量预测模型, 该模型具备较好的预测能力 ...

2017

0.0

... Li等^[6]利用红边区域特性, 提出了一种新的提取反射光谱的方法, 能很好的估算水稻和小麦叶片的叶绿素含量 ...

2015

0.0

... Lu等^[7]利用生长季节樱桃叶的多种光谱变量, 评价了高光谱遥感在估算叶绿素a、 b和叶绿素a+b浓度的能力, 实验结果的均方根误差较小 ...

1994

0.0

... Baret等^[8]建立光学辐射传输模型PROSPECT, 该模型定量描述了叶片色素、水分、结构参数等因素对叶片反射光谱的影响, 可以定量获取叶片内部各种成分的含量 ...

2018

0.0

... 例如, Jiang等^[9]比较植被指数与四个版本的PROSPECT模型反演性能, 发现两者性能类似, 但PROSPECT能够提供表面反射率和叶片内部结构等附加特性 ...

2018

0.0

... Sun等^[10]通过PROSPECT模型研究了采用32个通道的HSL系统估算叶片叶绿素含量的可能性, 结果表明PROSPECT模型适用于叶绿素含量的反演 ...

2014

0.0

... 然后扫描植物叶片的反射光R₁, 测量叶片的背景噪声B₂, 植物叶片的反射率光谱可表示为^[11] ...

2017

0.0

... 实验表明, 该模型版本优于之前版本, 能在可见光波段以最小的误差模拟真实叶片的光学性质, 在近红外波段、短波红外波段的误差和其他版本类似^[12] ...

2016

0.0

... 以X₀为参考点, 定义敏感度计算公式为^[13] ...

2015

0.0

... _X均以相等的相对间隔取值^[14] ...

2016

0.0

Hong-sheng

, YIN

Kai-le

(苏宏升, 殷凯乐). Computer Engineering and Applications(计算机工程与应用), 2016, 52(24): 50.

Considering that the existing Artificial Bee Colony（ABC） can not pay simultaneous attention to evolution speed and solution quality, a Nelder-Mead Simplex Method based Improved Artificial Bee Colony（NMSM-IABC） is proposed in this paper. In the process of iteration, the algorithm periodically gets the best individual vertex from NMSM operator and migrates to ABC, or obtains the optimal nectar source information from ABC and migrates to NMSM. ABC can improve its local exploiting capability applying NMSM, and NMSM can get away from local minimum by ABC. The algorithm proposed can achieve cooperative search of the ABC and NMSM. Furthermore, in order to enhance the ability or increase the convergence speed of NMSM-IABC, an improved search scheme of onlooker bee is proposed. And the sensitivity analysis of the key parameter is conducted. Finally, numerical experiments and comparisons on six benchmark functions indicate that the proposed algorithm can avoid the local minimum and enhance the global search ability and convergence speed, and is an effective cooperative search algorithm.

School of Automation and Electrical Engineering, Lanzhou Jiaotong University, Lanzhou 730070, China

针对现有的人工蜂群算法（Artificial Bee Colony，ABC）在进化速度和求解质量方面难以兼顾的缺点，提出一种基于Nelder-mead单纯形法的改进人工蜂群算法（Nelder-Mead Simplex Method based Improved Artificial Bee Colony，NMSM-IABC）。在迭代过程中，该算法周期性地将单纯形算子得到的最优个体迁移到人工蜂群算法的蜂群中，或将蜂群中的最优蜜源信息迁移到Nelder-mead单纯形算法中。旨在ABC借助NM-SM提高局部搜索能力，NM-SM借助ABC跳出局部最优点，达到两者协同搜索。再者，为了进一步加快收敛速度，在ABC中采用一种改进的跟随蜂搜索策略，并对产生侦察蜂的关键参数进行灵敏度分析。最后，通过6个典型的多维测试函数对算法进行仿真测试。结果表明：提出的算法有效地避免了陷入局部最优，提高全局搜索能力和搜索精度，有较快的收敛速度，是一种较好的协同搜索算法。

... 该算法是一种局部搜索算法, 在每次迭代过程中, 求出一个较好的探索点, 除去最差的点, 并用探索点来代替, 形成新的单纯形^[15] ...

2014

0.0

YANG

Ren-jie

, YANG

Yan-rong

, DONG

Gui-mei

, et al(杨仁杰, 杨延荣, 董桂梅, 等). Spectroscopy and Spectral Analysis(光谱学与光谱分析), 2014, 34(8): 2098.

Based on Euclidian distances between synchronous two-dimensional infrared correlation spectra, in terms of the average Euclidian distances between unknown samples and "extreme samples", and average intra- and inter- Euclidian distances of samples in the calibration set, a new method for the discrimination of adulterated milk was proposed. Sixteen pure milk samples were collected and 16 adulterated milk samples with urea (0. 01 similar to 0. 3 g . L-1), and 16 adulterated milk samples with melamine (0. 01 similar to 0. 3 g . L-1) samples were prepared, respectively. The IR absorption spectra of all samples were measured at room temperature. The synchronous two-dimensional correlation spectra were generated from concentration-dependent spectral variation of adulterant in milk. The Euclidian distances were calculated between synchronous two-dimensional infrared correlation spectra of all samples. Then, the classification models were built respectively for adulterated milk with urea, and adulterated milk with melamine. The "extreme samples", average intra- and inter- Euclidian distances were determined. Finally, the unknown samples in prediction set were predicted using constructed models in terms of classification rules of adulterated milk. The classification accuracy rates for pure milk and adulterated milk were 100%. The effectiveness of the proposed method was verified. The results obtained in this study revealed that synchronous two-dimensional infrared correlation spectra in combination with Euclidian distance has a feasible potential to discriminate adulterated milk and pure milk.

[Yang Ren-jie; Yang Yan-rong; Dong Gui-mei; Du Yan-hong; Shan Hui-yong; Zhang Wei-yu] Tianjin Agr Univ, Coll Engn & Technol, Tianjin 300384, Peoples R China.

... 其反映了空间中样本之间的不相似性^[16], 适用于计算本研究中光谱拟合的准确性 ...

2015

0.0

... 根据透过率光谱, 利用下式计算植物叶片的SPAD值^[17] ...

2009

0.0

2013

0.0