基于PLSR的苎麻叶片含水量估测模型建立及优化

为建立基于高光谱的苎麻叶片含水量估测模型,在大田栽培条件下,采集了360个苎麻叶片高光谱数据和相应的叶片含水量。用高杠杆值排除异常样本,用浓度梯度法划分样本集。采用多种光谱预处理方法,建立并比较各预处理方法的PLSR(partial least squares regression)模型效果,其中OSC(orthonormal signalcorrection)预处理方法最佳,预测集R^2=0. 8503,RMSEp=0. 0235。为了减少变量个数,通过OSC_PLSR模型的回归系数RC(regression coefficient)选择特征波段EB(effective bands)作为输入变量。随后,为了进一步降低计算量,本研究提出的一种新的特征提取方法:在基于RCEB建立的PLSR模型中,再次提取RC特征波长EW(effective wavelength)。由建模结果可知:与全波段相比,2种特征提取方法的变量个数均大幅减少(全波段为2 031个,RCEB为508个,RCEB_EW为16个); RCEB_PLS模型预测集指标最佳(R^2=0. 8546,RMSEp=0. 0232);与RCEB_PLS模型相比,RCEB_EW_PLSR模型预测集指标略低(R^2=0. 8499,RMSEp=0. 0234),但这种方法变量个数最少,因此综合评价效果最优。研究探讨了叶片高光谱与含水量之间的量化关系,建立基于高光谱的叶片含水量预测模型,对作物栽培中水分的实时监测和精确诊断具有实际指导意义。     

橡胶树叶片高光谱特征分析

从光谱曲线特征和光谱变换特征分析橡胶树(Hevea brasiliensis)叶片反射曲线特征.结果表明,蓝边、红边、黄边位置特征分别出现于525 nm、725 nm、550 nm波段附近,红谷位置特征变化较大,并提取了红边积分面积等重要光谱变量特征.叶片氮含量与反射光谱的相关分析表明橡胶树叶片氮素敏感波段为700～1300 nm,其中730 nm处相关性最好,达到0.8422的极显著水平,以730 nm处的反射率与叶片氮含量建立线性模型,其复相关系数R2达到0.7094.     

基于减量精细采样法估算小麦叶片氮积累量的最佳归一化光谱指数

基于6个小麦品种、5个施氮水平、4年田间试验条件下不同生育时期的小麦叶片高光谱反射率和相应的氮含量及生物量,采用减量精细采样法,系统构建了350～2500 nm范围内所有两两波段组成的归一化光谱指数［NDSI（i, j）］,综合分析了小麦叶片氮积累量(LNA, g N·m^-2)与NDSI（i, j）的定量关系,确定了估算叶片氮积累量的新高光谱特征波段和光谱指数,进而建立了小麦叶片氮积累量监测模型.结果表明：估算小麦叶片氮积累量的敏感波段主要存在于可见光区和近红外区,最佳特征波段组合为720 nm和860 nm;基于NDSI(860,720)的叶片氮积累量监测模型为LNA=26.34×［NDSI(860,720)］^1.887（R²=0.900,SE=1.327）.利用独立试验资料的检验结果表明,基于NDSI(860,720)建立的回归模型对小麦叶片氮积累量的估测精度为0.823,RMSE为0.991 g N·m^-2,模型预测值与观察值之间的符合度较高.可利用新的归一化高光谱参数NDSI(860,720)来估算小麦叶片氮积累量.     

秋茄类胡萝卜素含量高光谱反演

类胡萝卜素(Car)作为植物主要色素,对诊断植被生理状态有重要作用。于2013年4月和7月采集闽江口秋茄(Kandelia candel)叶片,室内测定其叶片正面和背面反射光谱,同时测定其Car含量\[单位面积(μg·cm-2)和单位质量(mg·g-1)\]。选取常见Car含量估算的光谱参数,同时分析确定最佳比值植被指数(SR),基于回归分析,建立秋茄叶片Car含量估算与验证模型。结果表明,叶片光谱反射率表现为叶片背面大于正面(350~2350 nm);基于叶面背面光谱计算的SR与叶片Car含量(μg·cm-2)的相关系数优于其他组合,相关系数较高的区域分布在520~540 nm与1000~1100 nm波段组合,700~720 nm与800~1100 nm波段组合;基于背面光谱计算的大部分光谱参数与Car含量(μg·cm-2)的相关系数要高于基于正面光谱计算的。因此,以叶片背面光谱作为Car含量估算的光谱数据,以单位面积Car含量为估算量纲建立反演模型。本研究表明,光谱指数LCI、DD、NDVI(770,713)、NDVI(773,562)、SR(723,770)和SR(1000,700)均可实现Car含量的反演,估算与检验模型的R2均0.65,RMSE均1.52;并且新构建的SR(1000,700)估算精度最好,模型和检验R2分别为0.77和0.87,模型和检验RMSE分别为1.08和1.11。这些预味着基于高光谱遥感对闽江河口湿地秋茄Car含量进行估算是可行的。     

小麦叶片氮素状况与光谱特性的相关性研究

 系统分析了不同时相下两个小麦(Triticum aestivium)品种叶片含氮量及叶片氮积累量与冠层光谱反射特征的关系。结果表明,随施氮水平的增加,小麦冠层在可见光区的反射率逐渐降低,而近红外波段的反射率逐渐升高。小麦叶片氮素状况与比值指数或归一化指数显著相关,两个品种表现极为一致,可以用一个指数方程来拟合。分阶段建模并没有提高模型的精度,因此可以建立一个适用于整个生育时期的通用氮素诊断方程。叶片含氮量同光谱指数在整个生育期内的关系要优于叶片氮积累量的,其中,与叶片含氮量关系最佳的指数为红波段(660 nm)和蓝波段(460 nm)的组合(R2>0.80);与叶片氮积累量关系最佳的光谱指数为中红外波段(1 220 nm)与红波段(660 nm)的组合(R2>0.62)。     

基于实测光谱的大兴安岭地区典型森林枯落物含水率估测模型

大兴安岭是我国重点森林火灾区,准确预测该地区的森林可燃物含水率对于提高该地区林火发生预测的准确性意义重大。本研究采集典型林型的枯落物的光谱和含水率实测数据,通过一阶导数和去包络线的光谱分析方法识别森林枯落物含水率敏感波段。通过相关系数法从原始光谱、去包络线光谱、一阶导数光谱、去包络线之后的一阶导数光谱中筛选与枯落物含水率高度相关的波段作为含水率反演模型的备选自变量。利用逐步回归分析建立枯落物含水率反演模型,并对模型进行精度评价。结果表明,去包络线之后的一阶导数光谱对枯落物含水率变化存在显著响应,敏感波段位于398~668、768~1068、1098、1278、1388~1438、1458~1538、1868~1898、1988~2088、2198~2208、2228~2238 nm(P0.05)。相关系数极值为-0.653、0.610,分别在波长2008、1888 nm处。通过多元逐步回归构建大兴安岭地区9种典型林型枯落物光谱和含水率的预测模型,模型决定系数R2=0.537,平均相对误差为0.303,均方根误差为0.499。本研究结果将为利用遥感技术快速测定森林枯落物含水率提供参考。     

棉花冠层高光谱参数与叶片氮含量的定量关系

建立棉花(Gossypium hirsutum)氮素状况的光谱监测技术对于棉花营养诊断和长势估测具有重要意义。该研究利用冠层高光谱反射率及演变的多种高光谱参数,分析了不同施氮水平下不同棉花品种叶片氮含量与冠层反射光谱的定量关系,建立了棉花叶片氮含量的敏感光谱参数及预测方程。结果显示,棉花叶片氮含量和冠层高光谱反射率随不同施氮水平呈显著变化。棉花叶片氮含量的敏感光谱波段为600~700 nm的可见光波段和750~900 nm的近红外波段,且叶片氮含量与比值植被指数RVI [average (760~850), 700]有密切的定量关系,4个品种的平均决定系数在0.70左右。进一步分析表明,可以用统一的回归方程来描述不同品种、不同生育时期和不同氮素水平下棉花叶片氮含量随反射光谱参数的变化模式,从而为棉花氮素营养的监测诊断与精确施肥提供技术依据。     

基于拉曼光谱和化学计量学方法判别大米分类的研究

本文利用拉曼光谱和化学计量学方法,建立快速分类模型对大米进行区分。在使用最小二乘法对离散拉曼光谱进行多项式拟合去除荧光背景的前提下,利用在第一次迭代过程去除大型拉曼峰和计算噪声电平的方法,并且保留数据维数在原来的50%以下。获取精确的拉曼信号。再用主成分分析法(Principal component Analysis,PCA)对3种大米全波段的拉曼光谱进行降维分析,线性判别方法 (Linear discrimination analysis,LDA)对样品进行分类,结果显示采用前两个主成分能达到93.8%的正确分类,采用前三个主成分能达到97.9%的正确分类。优化之后的模型对于大米的判别分析具有很好的效果。     

应用间接排序识别盐沼植被的光谱特征:以崇明东滩为例

遥感技术已成为大尺度植被分类的重要手段,而地面植物群落特征与其光谱特征之间的关系是解译遥感影像的关键。该研究选择上海崇明东滩自然保护区的盐沼植物群落为对象,应用ASD地物光谱仪测定其植物群落的光谱反射率,并采用10个小型机载成像光谱仪(CASI)默认植被波段组,应用主分量分析法和相关分析分析了不同群落光谱特征与生态环境因子之间的关系。分析结果表明,间接排序法PCA能够识别盐沼植被中光滩、海三棱 草(Scirpus mariqueter)群落、芦苇(Phragmites australis)群落和互花米草(Spartina alterniflora)等群落的光谱特征,绝大多数盐沼湿地植物群落组成与光谱特征之间有显著的相关,识别效果最好的波段组是736~744 nm、746~753 nm、775~784 nm、815~824 nm和860~870 nm;对光谱反射率影响最大的生态环境因子分别是植物群落的高度和盖度,高程和其它环境因子的影响次之。研究成果可为遥感监测崇明东滩自然保护区内入侵种互花米草的空间分布和扩散规律提供技术支撑,为高光谱遥感影像的影像判读和解译分类以及盐沼湿地植被制图提供科学依据。     

基于光谱参数对小白菜叶片镉含量的高光谱估算

为实现利用高光谱技术快速、准确、无损地检测叶类蔬菜叶片重金属镉污染情况,通过采用室内盆栽试验,检测了小白菜在6个不同的镉浓度梯度0 mg/kg(CK)、0.5 mg/kg(T1)、1 mg/kg(T2)、5 mg/kg(T3)、10 mg/kg(T4)和20 mg/kg(T5)下的叶片高光谱反射率及其镉含量。利用相关分析和逐步回归的统计方法对叶片原光谱、一阶导数光谱和光谱参数与镉含量进行统计分析,确定了反演叶片镉含量的敏感光谱参数,并建立了估算叶片镉含量的参数模型。结果表明:(1)在540 nm附近和红外区域,叶片光谱反射率随着处理浓度的增加呈下降趋势。T1组叶片光谱与对照组的光谱没有明显的变化差异;(2)原光谱与镉含量的敏感波段主要在690—1300 nm,相关系数最高的波段是782 nm。一阶微分光谱与镉含量的敏感波段在黄边、红外、近红外和远红外范围均有分布;(3)反映植物色素、水含量和细胞结构的参数MCARI(叶绿素吸收反射修正指数Modified Chlorophyll Absorption Reflectance Index),SDy(黄边面积Yellow Edge Area),WI(水质指数Water Index),DCWI(病态水分胁迫指数Disease Water Stress Index),SDr(红边面积Red Edge Area)和Dr(红边幅值The Amplitude of the Red Edge)可分别作为反演镉含量的敏感光谱参数,其倒数回归模型能够较好地反演镉污染下小白菜叶片的镉含量;(4)镉胁迫处理15 d时,建立的SDr的倒数模型估算处理30 d时小白菜叶片镉含量的效果最优。研究表明红边面积参数可以用于估算小白菜叶片的镉含量,可为评价小白菜的食用安全提供科学方法。     

Oil Adulteration Identification by Hyperspectral Imaging Using QHM and ICA

To investigate the feasibility of identification of qualified and adulterated oil product using hyperspectral imaging(HIS) technique, a novel feature set based on quantized histogram matrix (QHM) and feature selection method using improved kernel independent component analysis (iKICA) is proposed for HSI. We use UV and Halogen excitations in this study. Region of interest(ROI) of hyperspectral images of 256 oil samples from four varieties are obtained within the spectral region of 400–720nm. Radiation indexes extracted from each ROI are used as feature vectors. These indexes are individual band radiation index (RI), difference of consecutive spectral band radiation index (DRI), ratio of consecutive spectral band radiation index (RRI) and normalized DRI (NDRI). Another set of features called quantized histogram matrix (QHM) are extracted by applying quantization on the image histogram from these features. Based on these feature sets, improved kernel independent component analysis (iKICA) is used to select significant features. For comparison, algorithms such as plus L reduce R (plusLrR), Fisher, multidimensional scaling (MDS), independent component analysis (ICA), and principle component analysis (PCA) are also used to select the most significant wavelengths or features. Support vector machine (SVM) is used as the classifier. Experimental results show that the proposed methods are able to obtain robust and better classification performance with fewer number of spectral bands and simplify the design of computer vision systems.     

Growth characteristics of three Fusarium species evaluated by near-infrared hyperspectral imaging and multivariate image analysis

Colony growth of three Fusarium spp. on potato dextrose agar was followed by collecting near-infrared (NIR) hyperspectral images of the colonies at regular intervals after inoculation up to 55?h. After principal component analysis (PCA), two clusters were apparent in the score plot along principal component 1. Using the brushing technique, these clusters were divided into four groups of pixels with similar score values. These could be visualised as growth zones within the colonies in the corresponding score image. Three spectral bands, i.e. 1,166, 1,380 and 1,918?nm, were prominent in the multiplicative scatter corrected and Savitzky?CGolay second derivative spectra. These indicated chemical changes, associated with carbohydrates (1,166 and 1,380?nm) and protein (1,918?nm), that occurred as the mycelium grew and matured. The protein band was more prominent in the mature fungal material while the carbohydrate band was less pronounced. The younger material and the agar were characterised by the carbohydrate spectral band. Integrating whole mycelium colonies as the sum of pixels over time made it possible to construct curves that resembled growth curves; this included the lag phase, active growth phase, deceleration phase and phase of constant growth. Growth profiles constructed from individual growth zones indicated more detailed growth characteristics. The use of NIR hyperspectral imaging and multivariate image analysis (MIA) allowed one to visualise radial growth rings in the PCA score images. This would not have been possible with bulk spectroscopy. Interpreting spectral data enabled better understanding of microbial growth characteristics on agar medium. NIR hyperspectral imaging combined with MIA is a powerful tool for the evaluation of growth characteristics of fungi.     

基于“珠海一号”高光谱遥感影像的薇甘菊入侵监测

采用“珠海一号”卫星高光谱遥感影像,结合薇甘菊(Mikania micrantha)地面光谱数据,利用决策树分层分类提取方法,实现大范围薇甘菊入侵信息识别提取。结果表明,“珠海一号”OHS高光谱数据能准确反映薇甘菊精细光谱特征;在OHS高光谱数据的32个波段中,Band13 (665 nm)、Band22 (780 nm)、Band25 (833 nm)、Band26 (850 nm)、Band27 (865 nm)、Band28 (880 nm)是识别提取薇甘菊的最佳波段;利用OHS高光谱数据监测薇甘菊分布提取精度达82.6 %,可满足薇甘菊入侵监测日常业务需要,为薇甘菊精准防治提供有力支撑。     

水稻氮素营养高光谱遥感诊断模型

对水稻氮素含量与原始光谱反射率、一阶微分光谱以及高光谱特征参数间的相关性进行了分析,并构建和验证了以遥感参数为自变量的水稻氮素营养诊断模型.结果表明：氮素含量在水稻各器官中总的变化趋势为茎<鞘<穗<叶;各器官在可见光波段的光谱反射能力为叶<穗<鞘<茎,在近红外波段则与此相反.以波长796.7 nm处的光谱反射率和738.4 nm处的一阶微分光谱反射率为自变量的线性模型和指数模型的决定系数（R²）分别为0.7996和0.8606,二者均能较好地诊断水稻氮素营养,但最适合诊断水稻氮素含量的拟合模型是以植被指数的归一化变量（SD_r-SD_b）/（SD_r+SD_b）为自变量构建的水稻氮素营养高光谱遥感诊断模型［y=365871+639323（SD_r-SD_b）/（SD_r+SD_b）,R₂＝0.8755,RMSE＝0.2372,相对误差＝11.36％］,该模型可定量诊断水稻氮素营养.     

Gaussian band analysis of absorption,fluorescence and photobleaching difference spectra of D1/D2/cytb-559 complex

A study of the absorption and fluorescence characteristics of the D1/D2/cytb-559 reaction centre complex of Photosystem II has been carried out by gaussian decomposition of absorption spectra both at room temperature and 72 K and of the room temperature fluorescence spectrum. A five component fit was found in which the absorption and fluorescence sub-bands could be connected by the Stepanov relation. The photobleaching and light-activated degradation in the dark of long wavelength pigments permitted a further characterisation of the absorption bands. The absorption (fluorescence) maxima of the five bands at room temperature are 660 nm (670 nm), 669 nm (675 nm), 675 nm (681 nm), 680 nm (683 nm), 681 nm (689 nm). A novel feature of this analysis is the presence of two approximately isoenergetic absorption bands near 680 nm at room temperature. The narrower one (FWHM=12.5 nm) is attributed to pheophytin while the broader band (FWHM=23 nm) is thought to be P680. The P680 band width is discussed in terms of homogeneous and site inhomogeous band broadening. The P680 fluorescence has a large Stokes shift (9 nm) and most fluorescence in the 690–700 nm range is associated with this chromophore.The three accessory pigment bands are broad (FWHM=17–24 nm) and the 660 nm gaussian is largely temperature insensitive thus indicating significant site inhomogeneous broadening.The very slight narrowing of the D1/D2/cytb-559 Q_y absorption at crytogenic temperatures is discussed in terms of the coarse spectral inhomogeneity associated with the spectral forms and the apparently large site inhomogeneous broadening of short wavelength accessory pigments.     

基于挺水植物高光谱信息的再生水总氮含量估测——以北京市门城湖湿地公园为例

高光谱信息是探测植物体内氮素含量状况的重要手段,而植物体中的氮素与水体含氮量息息相关.本研究区为以再生水为主要补给水水源的北京门城湖湿地公园,通过获取区内典型的再生水氮净化挺水植物芦苇和香蒲叶片的高光谱数据,并在室内测定对应样点的水体总氮含量指标, 探讨基于典型湿地挺水植物高光谱数据对水体总氮进行遥感探测的可行性.采用4种高光谱参数(光谱指数、归一化差值指数、“三边”参数及吸收特征参数)分别建立一元线性模型、逐步多元回归模型和偏最小二乘模型,根据决定系数(R²)和均方根误差(RMSE)进行模型精度检验.结果表明: 逐步多元回归和偏最小二乘模型的预测精度高于一元线性模型. 3种模型对芦苇的拟合效果均优于香蒲.偏最小二乘模型对芦苇的拟合效果最优(R²=0.854,RMSE=0.647).500～700 nm是反映水氮含量的最佳波段范围,绿峰与红谷反射率的比值与水体总氮含量具有较强的相关性,尤其是吸收特征参数能够较好地预测水体总氮含量.     

基于因子分析的苜蓿叶片叶绿素高光谱反演研究

因子分析是一种能够将具有错综复杂关系的变量归结为少数几个综合因子的多变量统计分析方法,在降低数据维数的同时又可以保存足够的信息,这为处理信息量丰富但冗余较大的高光谱数据提供了一种有效方法。本文利用2010年9月23日采集的16个样点的苜蓿叶片反射率及叶绿素含量数据,采用因子分析方法,分别提取苜蓿叶片反射率光谱400～900nm,以及可见光400nm～760nm和近红外760～900nm光谱区的公共因子,分析因子载荷分布、载荷总量对公共因子与叶绿素含量相关性的影响。利用逐步回归法建立基于公共因子的叶片叶绿素反演模型,并将反演模型与光谱指数建立的模型进行对比。研究表明,1）公共因子与叶片叶绿素的相关性,在更大程度上是与该因子在各个波段上载荷分布有关,而不是总载荷量;2）对波谱进行分区建立的反演模型略优于全区因子分析建立的反演模型;3）与常用于叶片叶绿素含量反演的光谱指数CARI、MCARI、mND680、mND705、mSR705、TVI、DmSR、BGI、BRI相比,因子分析建立的叶绿素反演模型精度更高。     

Nonpigment components of the photochlorophyllide photoactive complex: studies of low-temperature blue-green fluorescence spectra

Fluorescence spectra in the blue-green region and excitation fluorescence spectra of green wheat leaves, etiolated wheat leaves and isolated inner etioplast membranes (prolamellar bodies and prothylakoids) were compared to specify the structure of the active protochlorophyllide pigment-protein complex of inner etioplast membranes. Three bands in the blue region at 420, 443 and 470 nm and a broader green band at 525 nm were found. Comparison of the emission and excitation spectra suggests that the main components responsible for the blue fluorescence of etioplast inner membranes are pyridine nucleotides and pterins. The green fluorescence (525 nm) excitation spectra of etiolated samples were identical to the excitation spectrum of flavin fluorescence. The fact confirms the suggestion that flavins are the constituents of the active protochlorophyllide-protein complex.     

Estimation algorithm for chlorophyll-a concentrations in water from hyperspectral images based on feature derivation and ensemble learning

Multicollinearity between feature bands is one of the main interferences in the process of retrieving chlorophyll-a (Chl-a) concentration in water bodies from hyperspectral data. Meanwhile, the model capability is also a decisive factor for inversion accuracy. To eliminate multicollinearity between feature bands and enhance data-driven of the retrieve of Chl-a concentration, this study proposed a feature bands selection strategy (FD-FI) based on knee-point-detection and variance inflation factor (VIF). Then, to realize model-driven, nine machine learning algorithms are combined to construct a MixModel, which was compared with other models. Chl-a concentration in Nansi Lake was estimated using “Zhuhai-No.1” remote sensing images and field measured data. The results show that the FD-FI strategy can effectively eliminate multicollinearity between bands or band combinations (VIF < 7). Using the same model, the strategy proposed in this study has a higher accuracy than existing strategies. In the five-fold cross-validation, XGBoost_FD-FI obtained the best performance with Coefficient of Determinatio (R²) and Root Mean Squared Error (RMSE) of 0.8351 and 6.6477 μg/L. In addition, combined with MixModel, the FD-FI strategy further improves the accuracy of Chl-a retrieval with R² = 0.8664 and RMSE = 5.7926 μg/L. When the model was applied to remote sensing images, the Chl-a spatial distribution obtained by the FD-FI strategy on the four models was the most consistent. MixModel is more sensitive to very high and low Chl-a concentrations, and its generalisation is more stable. Overall, this study provides an innovative approach for the selection of feature bands and model construction for Chl-a retrieval from inland lakes.     

FINE STRUCTURE OF THE SYNAPTIC DISCS SEPARATED FROM THE GOLDFISH MEDULLA OBLONGATA

Synaptic discs are structures localized in the club ending synapses on the Mauthner cell lateral dendrite of the goldfish medulla oblongata. The synaptic discs present a hexagonal array of particles ~8.5 nm center-to-center when observed in en face view. This lattice covers the entire surface Divalent cations are important in the stabilization of this particular hexagonal array of particles When a synaptic disc-rich fraction is treated with chelating agents (EDTA or EGTA), definite changes occur in the hexagonal lattice. First, the synaptic membranes show zones without particles interspersed with zones covered with the hexagonal array of particles Second, the synaptic discs break down and a new structure characterized by two parallel dense bands (7 nm each), separated by a 4 nm gap, is observed. The negative stain fills the gap region showing striations spaced ~10 nm center-to-center crossing the gap, but it does not penetrate the dense bands This "double band" structure is interpreted as an edge on view of a fragment of the synaptic membrane complex. Further treatment of this fraction with a chelating agent plus 0.3% deoxycholate produces an increase in the number of double band structures. However, EDTA plus Triton X-100 (a treatment known to produce solubilization of membrane proteins) never shows such double band structure An ordered material was observed associated with the cytoplasmic leaflets of the double bands This material consists of rows of beads ~4 nm in diameter and spaced at intervals of ~7 nm. Each of these beads is joined to the band by a thin stalk.