考虑品种差异的冷鲜猪肉含水率高光谱信号补正算法

为提高含水率高光谱校正模型对不同品种冷鲜猪肉样本的适用性,该文提出了1种品种敏感波段选择结合分段直接校正的高光谱信号补正方法。首先用主品种偏最小二乘回归模型和添加了7个从品种样本后更新模型的模型回归系数相对差异值,以此选出对品种敏感的波段,然后采用分段直接校正算法对选中波段的光谱信号补正,以消除品种差异对模型预测效果的影响。以猪肉含水率为检测指标,以零号土猪肉为主品种,以恩施山黑猪为从品种,经该算法补正后,主品种模型对从品种样本的预测均方根误差从补正前的1.42%减小到0.50%,预测决定系数从0.20提高到0.84,预测偏差比率(residual prediction deviation,RPD)从0.91提高到2.58。结果表明,该文提出的补正算法能显著提高零号土猪肉含水率模型对恩施山黑猪含水率的预测能力。     

高氧气调包装对不同品种冷却猪肉贮藏品质及持水性的影响

为了研究不同品种冷却猪肉在高氧气调包装贮藏过程中,蛋白氧化对猪肉品质及持水性的影响,试验以氟烷基因(NN)型杜长大三元杂交猪和不含NN基因型的三门峡黑猪为研究对象,分析了2个品种猪肉经高氧气调包装(80%O2+20%CO2),于(4±1)℃下贮藏过程中肌肉色泽、蛋白质氧化、水分分布、保水性以及肌肉微观结构的变化。结果表明,随着贮藏时间的延长(0~10 d),2个品种猪肉的肌原纤维蛋白羰基含量显著升高(P0.05),巯基含量显著降低(P0.05),表明猪肉肌原纤维蛋白的氧化程度随着贮藏时间的延长而加剧;肌原纤维蛋白发生持续性氧化,肌原纤维蛋白骨架的完整性遭到不同程度破坏,肌束膜破裂,纤维束间隙增大,结构疏松,保水性降低,不易流动水逐渐态变为自由水;与贮藏初始相比,第3天时杜长大三元杂交猪和三门峡黑猪的不易流动水均显著降低(P0.05),第5天杜长大三元杂交猪的自由水显著增加(P0.05),而黑猪的自由水在第10天时显著性增加(P0.05);贮藏5 d以上时,2个品种猪肉的蒸煮损失率均较对照组的蒸煮损失显著增大(P0.05);2个品种猪肉的L*值、a*值、b*值均呈现先增加再降低的趋势,L*值在第7天时达到最大值,a*值在第3天时达到最大值,杜长大三元杂交猪和黑猪的b*值分别在第5天和第7天时达到最大值;2个品种的猪肉经高氧气调包装,其色泽、蛋白质氧化、水分态变、保水性、微观结构等指标变化规律相似,表明高氧气调包装对2种猪肉的贮藏品质及持水性的影响效应具有一致性。研究结果为高氧气调包装冷却猪肉贮藏品质及汁液流失控制提供依据。     

手持式黄油桃可溶性固形物可见近红外光谱检测设备研制

桃在鲜果市场中占有重要份额。可溶性固形物含量（soluble solid content，SSC）是衡量桃品质的重要参数，是挑选优质桃以及预测最佳采摘时期的重要决策依据。该研究开发了一款基于可见近红外光谱技术的手持式黄油桃SSC无损检测设备。该设备的硬件系统主要由微型光谱仪、卤素灯、OLED显示屏、微控制器以及自主设计的驱动电路组成。为了评估所开发设备的检测性能，采用北京平谷区种植的黄油桃作为样品进行验证。首先，获取校正集样品在680～940 nm范围内的可见近红外光谱，经5点平均平滑和最大值归一化对光谱预处理建立黄油桃SSC偏最小二乘回归模型并用于预测集样本的SSC分析，预测相关系数和均方根误差分别为0.947和0.728%，单果检测时间不超过2 s。为了提高模型精度和稳定性，将校正集和预测集合并后作为新的校正集进行建模，并将重新构建的模型对独立验证集进行预测，SSC预测值与实测值的相关系数为0.906，均方根误差为0.732%。采用分段直接校正算法将主机模型传递到从机。经过模型传递后，从机对独立验证集SSC的预测值与实测值的相关系数和均方根误差分别为 0.865和0.919%。该手持式SSC检测设备可将SSC预测数据以蓝牙方式传输到手机客户端，借助手机定位功能，在地图上实现黄油桃SSC空间可视化分布。研究结果表明，该手持式SSC无损检测设备可以实现黄油桃SSC的准确测量，借助模型传递算法。实现了模型在不同设备间的有效传递，避免了重复建模，可为该设备批量生产节约大量成本，具有广阔的应用前景。     

近红外光谱和机器视觉信息融合的土壤含水率检测

为了精确、快速和稳定测定土壤含水率以及扩大所建模型的适应性,该文提出了机器视觉与近红外光谱技术融合的土壤含水率分析方法。通过试验建立了湖北地区主要土壤基于近红外光谱的土壤含水率分析模型、基于土壤表层图像特征参数的含水率分析模型和机器视觉与近红外光谱信息融合的土壤含水率分析模型。结果表明,基于近红外光谱含水率分析模型虽然具有较高的精度,但该模型预测非建模样品黄绵土误差均大于4%;以图像特征参数H,S和V所建BP人工神经网络非线性预测模型最优,模型的决定系数R2为0.9849,但当土壤水分饱和（达到20%以上）时存在分析误差;而所建立的土壤的近红外光谱与机器视觉BP神经网络信息融合模型可预测非建模样品黄绵土与水分饱和达20%以上土壤,决定系数R2可达到0.9961,融合模型分析精度均高于单独使用近红外光谱或机器视觉分析模型。     

基于高光谱的油麦菜叶片水分CARS-ABC-SVR预测模型

为了实现油麦菜生长期间更合理的灌水管理,研究一种基于高光谱技术的精确、快速、有效检测油麦菜叶片水分的新方法。以5种不同水分胁迫水平的油麦菜为研究对象,通过高光谱成像系统获取高光谱图像并利用干燥法测量叶片含水率。采用多项式平滑(Savitzky-Golay,SG)结合标准变量变换(standard normalized variable,SNV)对高光谱数据去噪平滑。利用竞争性自适应加权算法(competitive adaptive reweighted sampling,CARS)进行特征波长选择,并与逐步回归分析(stepwise regression,SR)及连续投影算法(successive projections algorithm,SPA)进行比较,利用支持向量回归机(support vector regression,SVR)分别建立油麦菜叶片全光谱数据、3种特征光谱数据与干基含水率的关系模型。结果表明,基于竞争性自适应加权算法波长选择的支持向量回归模型(CARS-SVR)效果最佳,但预测精度尚不够理想,故引入人工蜂群算法(artificial bee colony,ABC)优化模型的参数惩罚因子和核参数。最终,经人工蜂群算法优化后的模型(CARS-ABC-SVR)的预测集决定系数R2和均方根误差RMSE分别为0.9214和2.95%。因此,利用高光谱技术结合CARS-ABC-SVR模型预测油麦菜叶片水分含量是可行的。     

含水率对土壤有机质含量高光谱估算的影响

土壤含水率对有机质(soil organic matter,SOM)含量高光谱估算精度有很大的影响。为了探讨SOM高光谱估算中土壤含水率的影响,该文对烘干土、风干土和质量含水率为5%~40%(按5%递增)的土壤样本进行了室内高光谱测量,对光谱数据进行了反射率、反射率一阶导数和反射率倒数对数3种光谱数据变换,运用偏最小二乘回归法(partial least squares regression,PLSR)建立了相应的SOM估算模型。结果表明,风干土的SOM高光谱估算精度较好;当含水率水平小于25%时,SOM估算模型精度受含水率的影响较大,光谱数据进行反射率倒数对数变换后的模型精度最高;当含水率水平大于等于25%时,水分对土壤光谱反射率的影响要大于SOM,不适宜利用土壤光谱数据进行SOM含量高光谱估算。该研究可为大田环境不同含水率情况下光谱估算SOM提供参考。     

基于近红外光谱的板栗水分检测方法

含水率是影响板栗贮藏、加工的关键指标之一,该文应用近红外光谱技术对板栗含水率进行快速无损检测。试验对240个板栗样本的带壳光谱和栗仁板栗光谱采用SPXY算法进行样本集划分,利用偏最小二乘法建立含水率定量检测模型,并对微分、多元散射校正、变量标准化等多种预处理方法对建模结果的影响进行比较。结果表明：栗仁和带壳板栗的光谱经一阶微分预处理后所建模型性能最佳,其中栗仁的水分检测模型校正集和验证集的相关系数分别为0.9359和0.8473,校正均方根误差为1.44%,验证均方根误差为1.83%;带壳板栗光谱所建模型校正集和验证集的相关系数分别为0.8270和0.7655,校正均方根误差为2.27%,验证均方根误差为2.35%。受栗壳的影响,带壳板栗光谱模型对含水率的预测精度低于栗仁光谱模型的预测精度。研究表明,近红外光谱分析技术可用于板栗含水率的快速无损检测。     

黑土土壤水分高光谱特征及反演模型

应用高光谱技术阐释土壤含水率光谱规律及对土壤含水率进行定量分析,为精准农业地表土壤水分的快速测定提供参考。该文以吉林省黑土土类中的黑土亚类土壤为研究对象,利用ASD FieldSpec FR便携式光谱仪在室内环境下对不同含水率的土壤样本进行光谱反射率测量和特征分析;通过对土壤样品高光谱反射率进行对数、倒数、一阶微分以及反射率倒数的一阶微分、反射率对数的一阶微分变换,运用统计分析中的相关系数计算进行了光谱反射率与土壤水分的相关分析,并提取了土壤光谱特征波段;采用逐步多元线性回归方法和指数模式分析法,进行了高光谱土壤含水率定量反演。研究结果表明,在低于田间持水率状况下,黑土土壤光谱反射率及其反射率一阶微分和反射率对数一阶微分变换的敏感波段主要集中在400～410、1 400～1 850和2 050～2 200 nm范围内,其中2 156 nm处与土壤水分相关系数最高,达0.89;在波长1 328、1 439、1 742和2 156 nm处,采用反射率对数一阶微分所建立的黑土土壤含水率预测方程的预测精度最好,决定系数为0.931。黑土土壤含水率高光谱反演模型的建立为土壤水分的快速测定提供了新的途径。     

基于试验反射光谱数据的土壤含水率遥感反演

土壤含水率是土壤水循环研究中不可或缺的参数,已广泛应用于土壤水分的监测。土壤光谱特性的研究是土壤含水率光学遥感定量反演的基础。该研究首先通过野外调查收集土样;然后,在实验室条件下制备不同水分梯度的土壤样品,并利用便携式地物光谱仪采集不同水分梯度土壤样品的反射光谱;最后,通过试验光谱数据分析建立一个基于指数函数的土壤含水率遥感反演模型,并对结果进行精度评价。结果表明,基于指数函数的土壤含水率反演模型可以较好的反演土壤水分特征,在640 nm处土壤含水率的估计值与真实值之间的决定系数为0.7062,RMSE为3.49%。相关研究为表层土壤含水量的遥感监测提供新方法和新思路。     

基于冠层光谱特性的水稻叶片含水率模型

基于水稻叶片含水状况与冠层光谱反射率存在关联,尝试构建水稻叶片含水率模型。在水稻生长的孕穗期,同时测量室外水稻冠层光谱反射率和叶片含水率,依据水稻叶片含水率与各光谱波段反射率之间的相关性系数,选取高相关性系数对应的光谱特征波段。采用遗传算法对BP神经网络的初始权值进行优化处理。分别应用BP神经网络和GA-BP-Network、传统多元线性回归方法建立预测模型。试验表明,GA-BP-Network模型的预测含水率值与真实值平均误差率为3.9%,最大误差率为6.1%,均比BP神经网络、传统多元线性回归预测模型有了很大的改善,提高了预测水稻叶片含水率的准确性。     

土壤有机质含量可见-近红外光谱反演模型校正集优选方法

土壤有机质含量可见-近红外光谱反演过程中校正集的构建策略对模型的预测精度有重要影响。以江汉平原洪湖地区水稻土为研究对象,采用Kennard-Stone(KS)法,Rank-KS(RKS)和Sample set Partitioning based on joint X-Y distance(SPXY)法,构建样本数占总校正集不同比例的子校正集,通过偏最小二乘回归,建立土壤有机质含量的可见—近红外光谱反演模型。结果表明:KS法无法提高模型预测精度,但可以在保证标准差与预测均方根误差比(ratio of performance to standard deviation,RPD)2.0的前提下减少30%的校正样本;基于SPXY法的模型,当子校正集样本比例为总校正集的50%时达到最佳的模型预测精度,RPD为2.557;RKS法能够在保证预测精度的情况下(RPD2.0),最多减少总校正集70%的样本,对应模型RPD为2.212。当校正集与验证集的有机质含量分布相近时,能够以较少的建模样本达到与总校正集相近甚至更高的模型预测精度,提升土壤有机质光谱反演模型的实用性。     

高光谱技术结合CARS算法预测土壤水分含量

高光谱技术已成为预测土壤含水量(soil moisture content,SMC)的重要方法,但因土壤高光谱中包含了大量冗余信息和无效信息,不仅导致SMC的高光谱估算模型复杂度高,而且影响了模型的预测精度。因此,该研究在室内设计SMC梯度试验,测定土壤高光谱反射率,经Savitzky-Golay平滑(Savitzky-Golay smoothing,SG)和连续统去除(continuum removal,CR)预处理后,基于竞争适应重加权采样(competitive adaptive reweighted sampling,CARS)方法分别优选出土壤在全部SMC的水分敏感波长变量,确定适用于土壤在全部SMC的共性波长变量,以其为优选变量集,采用偏最小二乘(partial least squares regression,PLSR)回归方法建立模型并进行验证。结果表明,SG和CR预处理后的光谱曲线在450、1 400、1 900、2 200 nm附近吸收峰的形状特征凸显;基于CARS方法对土壤在不同SMC的光谱曲线进行变量优选后,得出优选变量集为443~449、1 408~1 456、1 916~1 943、2 209~2 225 nm;CARS-PLSR模型性能优于全波段PLSR模型,模型预测R2、均方根误差、相对分析误差分别为0.983、0.0144、8.36,不仅提升了预测精度和预测能力,而且降低了变量维度和模型复杂度。该文通过优选土壤水分的敏感波段,有效提高了SMC预测模型的鲁棒性,为快速准确评估农田墒情提供了新途径,为开发田间SMC测定传感器提供了理论依据。     

An Alternate Method for Fourier Transform Infrared (FTIR) Spectroscopic Determination of Soil Nitrate Using Derivative Analysis and Sample Treatments

This study aimed at examining effective sample treatments and spectral processing for an alternate method of soil nitrate determination using the attenuated total reflectance (ATR) of Fourier transform infrared (FTIR) spectroscopy. Prior to FTIR measurements, soil samples were prepared as paste to enhance adhesion between the ATR crystal and sample. The similar nitrate peak heights of soil pastes and their supernatants indicated that the nitrate in the liquid portion of the soil paste mainly responded to the FTIR signal. Using a 0.01-M CaSO₄ solution for the soil paste, which has no interference bands in the characteristic spectra of the analyte, increased the concentration of the nitrates to be measured. Second-order derivatives were used in the prediction model to minimize the interference effects and enhance the performance. The second-order derivative spectra contained a unique nitrate peak in a range of 1,400–1,200 cm^?1 without interference of carbonate. A partial least square regression model using second-order derivative spectra performed well (R ²?=?0.995, root mean square error (RMSE)?=?23.5, ratio of prediction to deviation (RPD)?=?13.8) on laboratory samples. Prediction results were also good for a test set of agricultural field soils with a CaCO₃ concentration of 6% to 8% (R ²?=?0.97, RMSE?=?18.6, RPD?=?3.5). Application of the prediction model based on soil paste samples to nitrate stock solution resulted in an increased RMSE (62.3); however, validation measures were still satisfactory (R ²?=?0.99, RPD?=?3.0).     

基于EPO算法去除水分影响的土壤有机质高光谱估算

野外进行土壤有机质的光谱快速预测时需考虑土壤含水量的影响。在室内设计人工加湿实验分别获取9个土壤含水量梯度(0~32%,间隔4%)的土壤光谱数据,分析土壤含水量变化对光谱的影响,再利用外部参数正交化法(external parameter orthogonalization,EPO)进行湿土光谱校正,并结合偏最小二乘回归和支持向量机回归分别建立土壤有机质预测模型。结果表明,土壤光谱反射率随着土壤含水量的增加呈非线性降低趋势,偏最小二乘回归模型的预测偏差比为1.16,模型不可用;经EPO算法校正后,各土壤含水量梯度之间的光谱差异性降低,能实现土壤有机质在不同土壤含水量梯度的有效估算,偏最小二乘回归和支持向量机回归模型的预测偏差比分别提高至1.76和2.15。研究结果可为田间快速预测土壤有机质提供必要参考。     

Feasibility of diode-array instruments to carry near-infrared spectroscopy from laboratory to feed process control

Near-infrared calibrations were developed for the instantaneous prediction of the chemical and ingredient composition of intact compound feeds. Two rather different instruments were compared (diode array vs grating monochromator). The grating monochromator was used in a static mode in the laboratory, whereas the diode-array instrumentbetter adapted to online analysiswas placed on a conveyor belt to simulate measurements at a feed mill plant. Modified partial least squares (MPLS) equations were developed using the same set of samples analyzed in the two instruments. Sample set 1 ( N = 398) was used to predict crude protein (CP) and crude fiber (CF), while sample set 2 ( N = 393) was used for the prediction of one macroingredient (sunflower meal, SFM) and one microingredient (mineral-vitamin premix, MVP). The standard error of cross-validation (SECV) and the coefficient of determination (R2) values for CF were better using the monochromator instrument. However, results obtained for CP, SFM, and MVP using the samples analyzed in the diode-array instrument showed similar or even greater accuracy than those obtained using samples analyzed in the grating monochromator. The excellent predictive ability [R2> 0.95; RPD (ratio of standard deviation to SECV) > 3] obtained for CP, CF, and SFM opens the way for the online use of NIRS diode-array instruments for surveillance and monitoring in the manufacture, processing, and marketing of compound feeds. R2, RPD, and SECV values for MVP showed similar performance for both instruments. Although RPD values did not reach the minimum recommended for quantitative analysis, results are encouraging for an ingredient present in feed compounds in such very low amounts.     

Prediction of Starch Amylose Content Versus Total Grain Amylose Content in Corn by Near-Infrared Transmittance Spectroscopy

A study was conducted to investigate methods of improving a near-infrared transmittance spectroscopy (NITS) amylose calibration that could serve as a rapid, nondestructive alternative to traditional methods for determining amylose content in corn. Calibrations were developed using a set of genotypes possessing endosperm mutations in single- and double-mutant combinations ranging in starch-amylose content (SAC) from -8.5 to 76%, relative to a standard curve. The influence of three factors were examined including comparing calibrations made against SAC versus grain amylose content (GAC), developing calibrations using partial least squares (PLS) analysis versus artificial neural networking (ANN), and using all samples in the calibrations set versus using progressively narrower ranges of SAC or GAC in the calibration set. Grain samples were divided into calibration and validation sets for PLS analysis while samples used in ANN were assigned to a training set, test set, and validation set. Performance statistics of the validation sets that were considered were the coefficient of determination (R), the standard error of prediction (SEP), and the ratio of the standard deviation of amylose values to the SEP (RPD), which were used to compare all NITS models. The study revealed an NITS prediction model for SAC (R = 0.96, SEP = 5.1%, RDP = 3.8) of similar precision to the best GAC model (R = 0.96, SEP = 2.7%, RPD = 3.5). Narrowing the amylose range of the calibration set generally did not improve performance statistics except for PLS models for SAC in which a decrease in SEP values was observed. In one model, the SEP improved while R and RPD remained constant (R = 0.94, SEP = 4.2%, RPD = 2.8) when samples with SAC values <20% were removed from the calibration set. Although the NITS amylose calibrations in this study are of limited precision, they may be useful when a rough screening method is needed for SAC. For example, NITS may be useful to detect severe contamination during transport and storage of specialty grains or to aid breeders when selecting for amylose content from large numbers of grain samples.     

Estimating Deoxynivalenol Content of Ground Oats Using VIS‐NIR Spectroscopy

The potential of VIS‐NIR spectroscopy as a rapid screening method for resistance of Fusarium‐inoculated oats to replace the costly chemical measurements of deoxynivalenol (DON) was investigated. Partial least squares (PLS) regression was conducted on second‐derivative spectra (400–2,350 nm) of 166 DON‐contaminated samples (0.05–28.1 ppm, mean = 13.06 ppm) with separate calibration and test set samples. The calibration set had 111 samples, and the test set had 55 samples. The best model developed had three PLS components and a root mean square error of prediction (RMSEP) of 3.16 ppm. The residual predictive deviation (RPD) value of the prediction model was 2.63, an acceptable value for the purpose of rough screening. Visual inspection and the VIS spectra of the samples revealed that high‐DON samples tended to be darker in color and coarser in texture compared with low‐DON samples. The second‐derivative spectra showed that low‐DON samples tended to have more water and fat content than high‐DON samples. With an RMSEP value of 3.16 and RPD of value of 2.63, it seems possible to use VIS‐NIR spectroscopy to semiquantitatively estimate DON content of oats and discard the worst genotypes during the early stages of screening.     

基于偏最小二乘投影的可见/近红外光谱猪肉综合品质分类

针对全波段光谱技术的生鲜猪肉综合品质快速无损分类存在光谱数据量大、样本数量较少时分类准确率较低等缺点。该文提出了一种基于偏最小二乘(partial least squares,PLS)投影分析算法和支持向量机的生鲜猪肉综合品质分类器。利用基于偏最小二乘投影分析算法对全波段光谱数据进行数据降维,选取了13个特征波长。利用粒子群优化算法优化支持向量机惩罚参数和径向基核函数参数,优化后二者最优为4.939和0.01。利用选取的特征波长和优化后的参数建立了生鲜猪肉综合品质支持向量分类器。研究结果表明,分类器对训练集中白肌肉(pale,soft and exudative,PSE)、正常肉(reddish-pink,firm and non-exudative,RFN)和黑干肉(dark,firm and dry,DFD)的回判识别率分别为为88.46%、94.11%和92.31%;测试集中PSE、RFN和DFD预测正确率分别为84.62%、94.11%和84.62%。该分类器满足模型简单、预测准确率高等优点,为生鲜猪肉综合品质在线分级提供参考。     

便携式生鲜猪肉多品质参数同时检测装置研发

针对农畜产品检测现场的需求,基于可见/近红外光谱检测技术和嵌入式系统,开发了灵活方便的猪肉品质无损检测装置。该装置利用卤素灯作为光源,新型光导探头和微型光谱仪采集肉样光谱信息,通过ARM(advanced RISC machines)控制处理器进行集中控制和数据的处理;在内嵌linux操作系统上,采用Qt开发工具,设计出人性化的交互界面,并将猪肉品质的检测结果输出到装置触摸屏上。为了建立多品质无损检测数学模型,获取了猪肉里脊在400~1 000 nm波长范围内的光谱数据,通过国标方法测得猪肉里脊主要品质参数颜色(L*、a*、b*)和p H值,采用标准正态变量变换(standard normalized variate,SNV)和Savitzky-Golay(S-G)平滑对光谱数据进行预处理,并结合理化数据建立偏最小二乘(partial least squares regression,PLSR)模型。用全交叉验证法选取PLSR建模的主成分数。p H值、L*、a*和b*的预测相关系数为0.88、0.90、0.97和0.97,预测标准差为0.19、1.77、1.17和0.63。通过现场试验表明,轻便式多品质无损检测装置具有较高的检测精度,满足于猪肉的颜色和p H值等品质参数检测的要求。