基于特征选择与特征提取融合的鸡蛋新鲜度光谱快速检测模型优化

为有效提高鸡蛋新鲜度检测效率、优化检测模型,本研究结合波长特征选择和特征提取方法各自的优点,对二者进行有效融合共同优化鸡蛋新鲜度检测模型。利用一阶微分对550～950?nm范围内鸡蛋的可见-近红外透射光谱数据进行预处理,考虑到冗余光谱信息对模型精度的影响,使用特征选择方法中的竞争性自适应重加权(competitive?adaptive?reweighted?sampling,CARS)算法融合非线性特征提取局部切空间排列(local?tangent?space?alignment,LTSA)算法最小化光谱无用信息,建立支持向量机回归(support?vector?regression,SVR)模型,结果表明单一使用CARS特征波长选择建立模型得到训练集交叉验证相关系数(Rcv)为0.880 5,交叉验证均方根误差(root?mean?square?error?of?cross?validation,RMSECV)为8.59,预测集相关系数(Rp)为0.888 9,预测集均方根误差(root?mean?square?error?of?prediction,RMSEP)为8.42,融合LTSA特征提取方法后得到Rcv为0.896 0,RMSECV为8.04,Rp为0.898 3,RMSEP为8.18,与CARS-SVR模型相比较,融合模型预测精度均有所提高,同时数据维数再次减少14个,进一步简化了预测模型。研究表明,将特征选择与特征提取二者融合共同应用于鸡蛋可见-近红外光谱数据,不仅提升了光谱检测效率,而且提高了鸡蛋新鲜度预测模型精度,可为鸡蛋新鲜度光谱检测模型优化提供参考依据。     

基于光谱技术的支持向量机判别牛肉新鲜度

目的 实现生鲜牛肉新鲜度等级的无损快速判别。方法 用可见/近红外光谱检测系统, 获取储存1~18 d的36块牛肉样品的400~1600 nm范围的光谱信息, 以挥发性盐基氮理化值为分类依据。用多元散射校正(MSC)、变量标准化(SNV)、SG平滑预处理方法处理光谱数据, 分别建立牛肉新鲜度的支持向量机分类模型。结果 MSC+SG预处理后所建立的分类模型预测能力最好, 训练集和测试集的回判识别率和预测识别率分别为96.30%、100%, 验证集的识别率为88.89%。结论 可见/近红外光谱结合支持向量机, 对牛肉新鲜度进行无损快速判别是可行的。     

可见/近红外光谱技术无损检测新鲜鸡蛋蛋白质含量的研究

鸡蛋是一种重要的食品,蛋白质是鸡蛋的主要营养成分。本研究利用可见近红外反射光谱技术无损检测新鲜鸡蛋的蛋白质含量。使用光谱仪获取新鲜鸡蛋在400~1100 nm波段范围内的漫反射光谱;分别使用多元散射校正(MSC)法和一阶导数法(1-D)对反射光谱进行预处理;对反射光谱、MSC处理光谱和1-D光谱,使用逐步回归法判别法选择最优波长组合,建立多元线性回归模型,使用全交叉验证法验证模型。结果表明,可见/近红外反射光谱经过多元散射校正后,确定的10个最优波长(400、403.16、407.9、714.6、715、715.58、970.4、970.75、973和974.45 nm)组合建立模型的校正和验证结果最好:选定模型的校正结果为R=0.92,SEC=0.42%;验证结果为Rcv=0.89,SECV=0.47%。研究表明可见/近红外反射光谱技术可以较好的预测新鲜鸡蛋的蛋白质含量,本研究可为可见近红外光谱技术在鸡蛋营养成分的快速检测提供一定的理论基础。     

玉露香梨可溶性固形物近红外漫透射光谱在线检测

应用近红外漫透射光谱技术探索玉露香梨可溶性固形物在线无损检测的可行性。358个试验样本被分成建模集和预测集(269∶89),分别用于建立模型和验证模型的预测能力。通过对玉露香梨样品近红外漫透射光谱分析发现,样品光谱在625,725,800nm处存在3个波峰,在673,765,825nm处存在3个波谷。通过对比不同预处理方法,发现漫透射近红外光谱分别经一阶微分、移动窗口平滑和多元散射校正组合预处理后建立的模型效果最好。结合组合预处理方法建立了偏最小二乘和偏最小二乘支持向量机预测模型,经比较,偏最小二乘支持向量机模型预测能力更强,模型预测均方根误差和相关系数分别为0.316%和0.949。对比发现主成分分析和径向基函数有利于提高最小二乘支持向量机模型的预测能力。试验结果表明采用近红外漫透射光谱技术结合最小二乘支持向量机算法,实现了玉露香梨可溶性固形物在线无损检测。     

基于可见-近红外光谱及随机森林的鸡蛋产地溯源

为了研究快速无损鉴别鸡蛋产地的可行性,利用可见-近红外光谱技术,采集4种湖北不同产地鸡蛋的透射光谱(500~900 nm),利用中心化、归一化、标准正态变量(SNV)、Savitzky-Golay平滑滤波(SG)和多元散射校正(MSC)、直接正交信号校正(Direct Orthogonal Signal Correction,DOSC)算法对光谱数据进行预处理,采用t分布式随机邻域嵌入(t-distributed stochastic neighbor embedding,t-SNE)、主成分分析(PCA)方法对预处理后的数据降维,并将降维后的数据分别输入极限学习机(extreme learning machine,ELM)和随机森林(random forest,RF),建立鸡蛋产地溯源模型。比较两种方法建立的模型,发现运用DOSC预处理及t-SNE提取的光谱特征信息建立的RF模型鉴别效果最好,训练集和预测集的鉴别正确率分别为100%和98.33%。研究结果表明基于可见-近红外光谱技术对鸡蛋产地溯源是可行的,为进一步研究与开发鸡蛋产地溯源便携式仪器提供技术支持。     

基于近红外光谱的生鲜猪肉新鲜度实时评估

目的 运用近红外光谱对生鲜猪肉新鲜度进行实时评估。方法 利用多通道可见近红外光谱系统, 获取了猪肉表面380~1080 nm波长范围内的漫反射光谱数据, 采用多元散射校正(MSC)和变量标准化(SNV)的预处理方法, 然后使用偏最小二乘回归建立猪肉新鲜度的预测模型, 进而对猪肉新鲜度进行评价。结果 采用变量标准化处理后的偏最小二乘回归模型相对比较稳定, 建模效果比较好。对挥发性盐基氮 (TVB-N)的验证集的相关系数达到0.91, 对pH值的验证集的相关系数达到0.93。最后利用该模型对猪肉新鲜度进行评定, 评定准确率达92.9%。结论 实验中运用多点的测量方式提高了近红外检测的精度和稳定性, 对于实时检测评估生鲜猪肉的新鲜度有很大的潜力。     

䥺Symbol`@@基于近红外光谱漫透射技术的苹果可溶性固形物含量在线检测

为研究苹果的内部品质,提高检测的速度和稳定性,将近红外光谱漫透射技术应用于在线检测研究,并采取偏最小二乘回归(PLSR)算法结合不同光谱预处理方法建立苹果内部的可溶性固形物含量(SSC)的定量模型。结果表明:采用一阶微分结合多元散射校正(MSC)预处理后的模型最稳定,校正集和预测集的标准差分别为0.17和0.39,校正集的相关系数也达到0.988 3。试验结果说明近红外光谱漫透射技术能够快速、无损地检测出苹果的可溶性固形物含量。     

基于近红外光谱漫透射技术的苹果可溶性固形物含量在线检测

为研究苹果的内部品质,提高检测的速度和稳定性,将近红外光谱漫透射技术应用于在线检测研究,并采取偏最小二乘回归(PLSR)算法结合不同光谱预处理方法建立苹果内部的可溶性固形物含量(SSC)的定量模型。结果表明:采用一阶微分结合多元散射校正(MSC)预处理后的模型最稳定,校正集和预测集的标准差分别为0.17和0.39,校正集的相关系数也达到0.988 3。试验结果说明近红外光谱漫透射技术能够快速、无损地检测出苹果的可溶性固形物含量。     

可见/近红外光谱技术无损检测新鲜鸡蛋pH及蛋白质的研究

本文利用可见/近红外光谱技术检测新鲜鸡蛋p H和蛋白质。分别采集新鲜鸡蛋在400~1000 nm和900~1700 nm波长范围的漫反射光谱,使用多元散射矫正(MSC)、标准正态变量变换(SNV)等光谱预处理技术,选择最佳的预处理方法,使用偏最小二乘法(PLS)建立p H和蛋白质模型并对其进行评价。结果表明,基于900~1700 nm波长范围的光谱获得的p H模型较好,其校正集相关系数为0.948,预测集相关系数为0.855;基于400~1000 nm波长范围的光谱获得的蛋白质模型较好,其校正集相关系数为0.927,预测集相关系数为0.906。研究表明,可见/近红外光谱技术可以较好的预测新鲜鸡蛋的p H和蛋白质,为鸡蛋营养成分的快速无损检测提供新的思路和方法。     

基于高光谱成像的滩羊肉菌落总数和挥发性盐基氮的无损检测

以宁夏滩羊肉为研究对象,利用400~1000 nm可见近红外高光谱对冷鲜羊肉的菌落总数和挥发性盐基氮含量进行新鲜度的检测研究。采集冷鲜滩羊肉表面光谱图像,提取感兴趣区域获取原始光谱数据。剔除由蒙特卡洛检测法所检测出的异常样本,采用理化值共生距离法(SPXY)划分样本的校正集和预测集。先对原始光谱预处理并建立偏最小二乘回归(PLSR)模型,优选最佳预处理方法;后采用主成分回归法(PCR)和支持向量机回归法(SVR)建立模型,优选最佳建模方法。结果表明:光谱数据经过正交信号校正(OSC)预处理后,建立的菌落总数和TVB-N含量预测模型效果较好,其RC分别为0.9067和0.9147,Rp分别为0.8743和0.8802,均高于其他光谱预处理模型。通过不同建模方法的比较,建模效果较好的是PLSR方法。研究表明:利用可见近红外高光谱技术可以实现对滩羊肉新鲜度的无损检测。     

基于可见-近红外高光谱技术对鸡种蛋性别鉴定的研究

目的 为实现鸡种蛋胚胎性别的无损检测,提出了基于可见-近红外高光谱检测海兰褐鸡种蛋胚胎性别的方法。方法 通过分析种蛋0~14 d大头部位的400~1000 nm波段下的光谱,建立基于偏最小二乘判别分析（PLS-DA）和支持向量机（SVM）的种蛋性别判别模型,比较不同孵育天数下的模型判别率,优选出最佳的检测天数;通过分析四种不同的预处理算法,选出最佳的鸡种蛋胚胎高光谱预处理方法,最后构建基于全波段和特征波段光谱信息的判别模型,并对结果进行比较。结果 基于PLS-DA和SVM的模型在第9 d的预测集结果达到最高,分别为80%和82.5%。主成分分析（PCA）结果表明,雄雌种蛋光谱信息可以进行区分;变量标准化（SNV）为最佳预处理方法;全波段相对于连续投影算法（SPA）、竞争性自适应重加权算法（CARS）选择特征波长的模型更优,建模集、预测集准确率分别为90%和85%。结论 研究结果表明可见-近红外高光谱技术可以快速、较准确、无损检测海兰褐种蛋胚胎性别,该技术为褐壳种蛋胚胎性别鉴定实现在线检测提供了一定的理论基础。     

A Non-invasive Method for Screening Sodium Hydroxymethanesulfonate in Wheat Flour by Near-Infrared Spectroscopy

A rapid and non-invasive method, based on near infrared diffuse reflectance spectroscopy, was established for screening sodium hydroxymethanesulfonate in wheat flour. Successive projection algorithm was used for spectral variable selection. The selected variables were applied as inputs to partial least square discriminant analysis (PLS-DA) and advanced K-means dynamic clustering. The first two principal components extracted by PLS-DA had been applied as inputs to least squares support vector machine (LS-SVM). Three algorithms, including PLS-DA, advanced K-means dynamic clustering, and LS-SVM, were used to establish the calibration model. The results of LS-SVM outperformed that of the other two methods, with the classification accuracy of 92.0% for the validation and 94.7% for the prediction. The results of the study showed the potential of near-infrared spectroscopy as a non-invasive and environmentally acceptable method for the screening of sodium hydroxymethanesulfonate in wheat flour.     

基于高光谱图像与光谱特征融合技术的鸡蛋新鲜度无损判别模型的建立

鸡蛋新鲜度等级评价是鸡蛋品质检测过程中的一项重要技术指标。选取了不同储藏环境的鸡蛋样本并采集其高光谱图像信息与光谱信息,提取图像特征和光谱特征;采用并行式融合方法进行图谱特征融合,基于连续投影法-灰度共生矩阵方法进行特征提取;建立支持向量机鸡蛋新鲜度判别模型。采用粒子群算法优化模型,训练集准确率达到85%,预测集准确率达到76.67%。为了解决单模型可能出现的偶然性误判问题,采用递进式特征融合方法,引入多模型共识策略和深度残差网络ResNet 50分析方法。建立基于连续投影法-方向梯度直方图特征提取方法的多模型共识策略,该模型的训练集准确率提升至89%,预测集准确率提升至88%;同时,建立基于连续投影法-方向梯度直方图特征提取方法的深度残差网络ResNet 50模型,模型的训练集准确率提升至89%,预测集的准确率提升至86.67%。图谱特征融合建模分析表明,并行式融合方法和递进式融合方法对鸡蛋新鲜度等级判别都有一定的可识别性,且递进式融合算法的多模型共识策略判别效果更佳。     

Development of a rapid method based on front-face fluorescence spectroscopy for the monitoring of egg freshness: 2—evolution of egg yolk

This preliminary study is devoted to the application of front-face fluorescence spectroscopy to the study of egg yolks during storage. A total of 79 eggs stored for 1, 2, 3, 4, 5, 9, 10, 12, 16, 18, 23, 25 and 29 days at room temperature were analysed. The fluorescence emission spectra of tryptophan residues (excitation: 290 nm; emission: 305–430 nm) of proteins and the excitation spectra of vitamin A (emission: 410 nm; excitation: 270–350 nm) were recorded directly on egg yolk samples. Factorial discriminant analysis (FDA) was used to classify the eggs according to their date after they were laid. Using tryptophan fluorescence spectra, correct classification was observed for 57.1 and 51.9% for the calibration and the validation sets, respectively. Better classification (94.9 and 91.4% of the calibration and validation samples, respectively) was obtained from the vitamin A fluorescence spectra. The first five principal components (PCs) of the principal component analysis (PCA) extracted from each data set (tryptophan and vitamin A fluorescence spectra) were pooled (concatenated) into a single-matrix and analysed by FDA. Correct classifications were obtained for 97.5% of the calibration and 96.3.1% of the validation spectra. The discrimination of the investigated egg yolks according to their storage time was excellent. It was concluded that the concatenation of different fluorescence spectra might be considered as a promising indicator of shell egg freshness when they are used in egg products.     

Prediction of Egg Freshness and Albumen Quality Using Visible/Near Infrared Spectroscopy

Important changes occur in egg during storage leading to loss of quality. Prediction of these changes is critical in order to monitor egg quality and freshness. The aim of this research was to evaluate application of visible (VIS) and near infrared (NIR) spectroscopy as a rapid and non-destructive technique for egg quality assessment. Three hundred and sixty intact white-shelled eggs freshly laid by the same flock of hens fed with a standard feed were obtained. They were put under controlled conditions of temperature and humidity (T = 18 °C and RH = 55%) for 16 days of storage. Forty eggs were analyzed at day 0, 2, 4, 6, 8, 10, 12, 14, and 16. Transmission spectral data was obtained in the range from 350 to 2,500 nm. The non-destructive spectral data was compared to egg sample’s Haugh unit (HU) and albumen pH in terms of quality and to the number of storage days in terms of freshness. A partial least squares predictive model was developed and used to link the destructive assessment methods and the number of storage days with the spectral data. The correlation coefficient between the measured and predicted values of HU, albumen pH, and number of storage days were up to 0.94, R ² was up to 0.90 and the root mean square error values for the validation were 5.05, 0.06, and 1.65, respectively. These results showed that VIS/NIR transmission spectroscopy is a good tool for assessment of egg freshness and albumen pH and can be used as a non-destructive method for the prediction of HU, albumen pH, and number of storage days. In addition, the relevant information about these parameters was in the VIS and NIR ranging from 411 to 1,729 nm.     

鸡蛋新鲜度指标与贮藏天数相关性研究

目的 分析鸡蛋新鲜度指标与贮藏天数的相关性, 确定新鲜度指标数值显著变化的时间, 建立新鲜度与贮藏时间相关性模型。方法 2 d为一个时间点, 设立12个实验组, 每组5枚鸡蛋, 按照标准中鸡蛋新鲜度评定指标重复3次测定鸡蛋相对密度、气室高度、哈夫单位、蛋黄指数、pH值、失重率。根据Pearson相关系数评价贮藏天数与鸡蛋新鲜度各指标之间的相关程度, Duncan法评价贮藏天数与鸡蛋新鲜度各指标之间的差异显著程度。结果 鸡蛋新鲜度各指标与贮藏天数之间显著相关(P<0.05), 各新鲜度指标数值出现显著变化的时间各不相同, 各天数之间的鸡蛋失重量、失重率差异显著(P<0.05)。结论 本研究所建立的鸡蛋新鲜度与贮藏时间的相关性模型, 可作为鸡蛋新鲜度评价模型。     

Prediction of Poultry Egg Freshness Using Vis-Nir Spectroscopy with Maximum Likelihood Method

Crucial physio-chemical changes occuring in eggs during storage after laying lead to loss of egg freshness. In this research, a new method for prediction of egg freshness using transmission visible near infrared spectroscopy was investigated. For this purpose 300 eggs were stored at two control conditions: refrigerator (4–5°C, 75%RH) and room (24–25°C, 40%RH) then by special egg holder, transmission spectroscopy was measured. For two eggs groups, 25 eggs in each group, in six days were tested by spectroscopy, after that Haugh unit and air cell height was measured directly. The non-destructive visible near infrared spectroscopy spectral measurements from 300 to 1100 nm (832 length of wave) were done as well as Haugh unit, air cell height for each egg and created the database for both environments. Finally a maximum likelihood latent root regression algorithm was developed to predict Haugh unit and air cell height by spectrum observation. The database was randomly divided into two parts. Training data, was used for maximum likelihood latent root regression parameter tuning and training of the model and testing data, was used just for model evaluation. Results indicated that maximum likelihood latent root regression method showed good prediction ability with coefficient of determination (R²) value up to 0.82 and 0.86 for Haugh unit and air cell height, respectively for testing data set. The results showed this method was better in comparison with partial least square regression (R² up to 0.79 and 0.72 for air cell height and Haugh unit) which was already used for this prediction.     

基于近红外光谱的桃果实冷害识别分析

该文采用近红外(near infrared,NIR)光谱技术对水蜜桃低温冷害褐变进行识别分析。分别建立了水蜜桃低温贮藏期间不同冷害阶段的两分类和多分类模型,讨论了不同光谱预处理方法对模型的影响,并比较偏最小二乘判别分析(partial least squares-discriminant analysis,PLS-DA)、主成分判别分析(principal component discriminant analysis,PCA-DA)、K-最邻近(K-nearest neighbor,K-NN)、簇类独立软模式(soft independent modeling of class analogy,SIMCA)4种建模方法的分类效果。结果表明,采用PLS-DA模型效果最好,两分类和多分类模型的总准确率为分别为0.93和0.71。两分类模型可较准确地对冷害褐变进行快速识别分类,多分类模型可用于水蜜桃低温贮藏期间不同冷害阶段的初步筛查。     

基于卷积神经网络模型的裂纹鸡蛋图像识别

目的:提高基于计算机视觉的鸡蛋裂纹检测方法的准确性和运行效率。方法:使用禽蛋模拟撞击设备得到裂纹鸡蛋,并通过鸡蛋动态图像采集设备采集不同角度裂纹鸡蛋和完好鸡蛋图像,然后以原始图像和经预处理后图像分别建立用于裂纹鸡蛋图像识别的YOLO-v5、ResNet和SuffleNet模型,并比较不同模型识别准确度以及对未经预处理图像的适应性。结果:YOLO-v5、ResNet和SuffleNet模型均可有效识别经过预处理的裂纹鸡蛋图像,其验证集准确率分别为98.8%,97.8%,99.4%。对于未经预处理的裂纹鸡蛋,ResNet模型判别准确率较低,而SuffleNet模型对其适应性较好,判别准确度超过99%。结论:在卷积神经网络模型中,SuffleNet模型适用于裂纹鸡蛋图像的识别,且采集的图像无需进行预处理。