Visible and Near-Infrared Spectroscopic Discriminant Analysis Applied to Brand Identification of Wine

High-end wine brand is made through the use of high-quality grape variety and yeast strain, and through a unique process. Not only is it rich in nutrients, but also it has a unique taste and a fragrant scent. Brand identification of wine is difficult and complex because of high similarity. In this paper, visible and near-infrared (NIR) spectroscopy combined with partial least squares discriminant analysis (PLS-DA) was used to explore the feasibility of wine brand identification. Chilean Aoyo wine (2016 vintage) was selected as the identification brand (negative, 100 samples), and various other brands of wine were used as interference brands (positive, 373 samples). Samples of each type were randomly divided into the calibration, prediction and validation sets. For comparison, the PLS-DA models were established in three independent and two complex wavebands of visible (400 - 780 nm), short-NIR (780 - 1100 nm), long-NIR (1100 - 2498 nm), whole NIR (780 - 2498 nm) and whole scanning (400 - 2498 nm). In independent validation, the five models all achieved good discriminant effects. Among them, the visible region model achieved the best effect. The recognition-accuracy rates in validation of negative, positive and total samples achieved 100%, 95.6% and 97.5%, respectively. The results indicated the feasibility of wine brand identification with Vis-NIR spectroscopy.

Identification of Salmonella and Listeria monocytogenes by Near-infrared Spectroscopy and PCA

Near-infrared spectra of pathogenic bacteria (salmonella and Listeria monocytogenes ) were determined, and the spectral data were analyzed by the projection discriminant analysis based on principal component analysis (PCA). The expected results were obtained. The results showed that salmonella and L. monocytogenes could be distinguished from each other by the near-infrared spectroscopy of the whole cells, cell walls or cytoplasm.

Near-Infrared Spectroscopy Combined with Partial Least Squares Discriminant Analysis Applied to Identification of Liquor Brands

The identification of liquor brands is very important for food safety. Most of the fake liquors are usually made into the products with the same flavor and alcohol content as regular brand, so the identification for the liquor brands with the same flavor and the same alcohol content is essential. However, it is also difficult because the components of such liquor samples are very similar. Near-infrared (NIR) spectroscopy combined with partial least squares discriminant analysis (PLS-DA) was applied to identification of liquor brands with the same flavor and alcohol content. A total of 160 samples of Luzhou Laojiao liquor and 200 samples of non-Luzhou Laojiao liquor with the same flavor and alcohol content were used for identification. Samples of each type were randomly divided into the modeling and validation sets. The modeling samples were further divided into calibration and prediction sets using the Kennard-Stone algorithm to achieve uniformity and representativeness. In the modeling and validation processes based on PLS-DA method, the recognition rates of samples achieved 99.1% and 98.7%, respectively. The results show high prediction performance for the identification of liquor brands, and were obviously better than those obtained from the principal component linear discriminant analysis method. NIR spectroscopy combined with the PLS-DA method provides a quick and effective means of the discriminant analysis of liquor brands, and is also a promising tool for large-scale inspection of liquor food safety.

Identification of Plant-Pathogenic Fungi Using Fourier Transform Infrared Spectroscopy Combined with Chemometric Analyses

Identification of plant-pathogenic fungi is time-consuming due to cultivation and microscopic examination and can be influenced by the interpretation of the micro-morphological characters observed.The present investigation aimed to create a simple but sophisticated method for the identification of plant-pathogenic fungi by Fourier transform infrared(FTIR)spectroscopy.In this study,FTIR-attenuated total reflectance(ATR)spectroscopy was used in combination with chemometric analysis for identification of important pathogenic fungi of horticultural plants.Mixtures of mycelia and spores from 27fungal strains belonging to nine different families were collected from liquid PD or solid PDA media cultures and subjected to FTIR-ATR spectroscopy measurements.The FTIR-ATR spectra ranging from 4 000to 400cm-1 were obtained.To classify the FTIRATR spectra,cluster analysis was compared with canonical vitiate analysis(CVA)in the spectral regions of3 050~2 800and 1 800~900cm-1.Results showed that the identification accuracies achieved 97.53%and99.18%for the cluster analysis and CVA analysis,respectively,demonstrating the high potential of this technique for fungal strain identification.

Visible and Near-Infrared Spectroscopic Discriminant Analysis Applied to Identification of Soy Sauce Adulteration

The identification of soy sauce adulteration can avoid fraud, and protect the rights and interests of producers and consumers. Based on two measurement models (1 mm, 10 mm), the visible and near-infrared (Vis-NIR) spectroscopy combined with standard normal variate-partial least squares-discriminant analysis (SNV-PLS-DA) was used to establish the discriminant analysis models for adulterated and brewed soy sauces. Chubang soy sauce was selected as an identification brand (negative, 70). The adulteration samples (positive, 72) were prepared by mixing Chubang soy sauce and blended soy sauce with different adulteration rates. Among them, the “blended soy sauce” sample was concocted of salt water (NaCl), monosodium glutamate (C5H10NNaO5) and caramel color (C6H8O3). The rigorous calibration-prediction-validation sample design was adopted. For the case of 1 mm, five waveband models (visible, short-NIR, long-NIR, whole NIR and whole scanning regions) were established respectively;in the case of 10 mm, three waveband models (visible, short-NIR and visible-short-NIR regions) for unsaturated absorption were also established respectively. In independent validation, the models of all wavebands in the cases of 1 mm and 10 mm have achieved good discrimination effects. For the case of 1 mm, the visible model achieved the optimal validation effect, the validation recognition-accuracy rate (RARV) was 99.6%;while in the case of 10 mm, both the visible and visible-short-NIR models achieved the optimal validation effect (RARV = 100%). The detection method does not require reagents and is fast and simple, which is easy to promote the application. The results can provide valuable reference for designing small dedicated spectrometers with different measurement modals and different spectral regions.

Identification of Lubricating Oil Additives Using XGBoost and Ant Colony Optimization Algorithms

To address the problem of identifying multiple types of additives in lubricating oil,a method based on midinfrared spectral band selection using the eXtreme Gradient Boosting(XGBoost)algorithm combined with the ant colony optimization(ACO)algorithm is proposed.The XGBoost algorithm was used to train and test three additives,T534(alkyl diphenylamine),T308(isooctyl acid thiophospholipid octadecylamine),and T306(trimethylphenol phosphate),separately,in order to screen for the optimal combination of spectral bands for each additive.The ACO algorithm was used to optimize the parameters of the XGBoost algorithm to improve the identification accuracy.During this process,the support vector machine(SVM)and hybrid bat algorithms(HBA)were included as a comparison,generating four models:ACO-XGBoost,ACO-SVM,HBA-XGboost,and HBA-SVM.The results showed that all four models could identify the three additives efficiently,with the ACO-XGBoost model achieving 100%recognition of all three additives.In addition,the generalizability of the ACO-XGBoost model was further demonstrated by predicting a lubricating oil containing the three additives prepared in our laboratory and a collected sample of commercial oil currently in use。

Rapid, Non-Destructive, Textile Classification Using SIMCA on Diffuse Near-Infrared Reflectance Spectra

Soft independent modeling of class analogy (SIMCA) was successful in classifying a large library of 758 commercially available, non-blended samples of acetate, cotton, polyester, rayon, silk and wool 89% - 98% of the time at the 95% confidence level (p = 0.05 significance level). In the present study, cotton and silk had a 62% and 24% chance, respectively, of being classified with their own group and also with rayon. SIMCA correctly identified a counterfeit “silk” sample as polyester. When coupled with diffuse NIR reflectance spectroscopy and a large sample library, SIMCA shows considerable promise as a quick, non-destructive, multivariate method for fiber identification. A major advantage is simplicity. No sample pretreatment of any kind was required, and no adjust-ments were made for fiber origin, manufacturing process residues, topical finishes, weave pattern, or dye content. Increasing the sample library should make the models more robust and improve identification rates over those reported in this paper.

Manufacturer identification and storage time determination of “Dong'e Ejiao” using near infrared spectroscopy and chemometrics

We have developed a set of chemometric methods to address two critical issues in quality control of a precious traditional Chinese medicine （TCM）, Dong＇e Ejiao （DEE J）. Based on near infrared （NIR） spectra of multiple samples, the genuine manufacturer of DEE J, e.g. Dong＇e Ejiao Co., Ltd., was accurately identified among 21 suppliers by the fingerprint method using Hotelling T2, distance to Model X （DModX）, and similarity match value （SMV） as dis- criminate criteria. Soft independent modeling of the class analogy algorithm led to a misjudgment ratio of 6.2%, suggesting that the fingerprint method is more suitable for manufacturer identification. For another important feature related to clinical efficacy of DEE J, storage time, the partial least squares-discriminant analysis （PLS-DA） method was applied with a satisfactory misjudgment ratio （15.6%） and individual prediction error around 1 year. Our results demonstrate that NIR spectra comprehensively reflect the essential quality information of DEE J, and with the aid of proper chemometric algorithms, it is able to identify genuine manufacturer and determine accurate storage time. The overall results indicate the promising potential of NIR spectroscopy as an effective quality control tool for DEEJ and other precious TCM products.

基于红外光谱与聚类分析法的宁夏产地枸杞子品种鉴别研究

宁夏枸杞(Lycium barbarum L.)为宁夏传统道地中药材,因其良好的药食两用价值,已引种至甘肃、新疆、青海等省区。因长期引种和改良,宁夏枸杞品种繁多,是影响其质量的重要因素之一。本研究采用傅里叶变换红外光谱(FTIR)对宁夏中宁产地的宁杞1号、4号、5号、7号、0909号、10号共6个品种的枸杞子进行测定,一维红外光谱及二阶导数光谱范围均为4000~650cm^(-1),得到图谱后进行图谱解析;采用谱带较密集的指纹区(1800~650cm^(-1))计算红外光谱图的相似系数;然后结合SMICA聚类分析法对不同品种枸杞子的红外指纹图谱进行聚类比较。结果表明,不同品种枸杞子样品的一维红外光谱图较为相似,峰的位置、峰高和峰形都较为接近,其共同吸收峰较多,仅在3282~3288、1239~1242和1143~1147 cm^(-1)附近的吸收峰强度、峰位置、峰形状有所不同,说明不同品种枸杞子中多糖类、苷类、蛋白质类、脂类和黄酮类等成分的种类和含量有所差异。二阶导数光谱中,宁杞1号、4号、7号在2880 cm^(-1)处吸收峰不明显,0909号在969 cm^(-1)处吸收峰不明显。不同品种枸杞子的相似度范围在0.9489~0.9928之间,说明不同品种枸杞子存在一定差异,其中,宁杞7号与其他品种的平均相似度系数最小,为0.9640,说明其成分特异性最高。0909号与宁杞10号相似系数为0.9928,相似度最高。采用Assure ID软件以各药材吸收波数为变量进行聚类分析,宁杞1号与其他品种的差异均较小,类间距在2.17~2.97范围内;0909号与其他品种间的差异最大,类间距在2.97~8.06范围内。聚类模型中,不同品种枸杞子的识别率均为100%,仅宁杞1号的拒绝率较低,为66%,易与其他品种枸杞子混淆,0909号的拒绝率和识别率均为100%,最易区分。类模型图中,0909号与宁杞5号、0909号与宁杞7号两两分开,可以明显鉴别出不同品种的样品。采用已知品种的枸杞子对聚类分析模型进行验证,其识别率和拒绝率与聚类分析模型相同。不同品种枸杞子类模型图与一维红外光谱、二阶导数图谱、相似度、类间距、识别率和拒绝率结果相互印证,佐证了宁杞1号为大多数宁杞系列品种的来源品种,而09系列品种0909号与其他品种有较大差异。因此,红外光谱与聚类分析法相结合可以快速、无损地鉴别不同品种枸杞子,对枸杞子药材生产和新品种选育具有一定的指导作用。

基于红外光谱技术智能识别润滑油的研究进展

机器学习作为人工智能发展的核心,在各行业得到快速发展,近年来也成为润滑油领域研究的热点之一,标志着润滑油的研究不再局限于大规模的试验研究,高通量数据、机器学习、优化算法开始应用于润滑油的研究。文章介绍了基于红外光谱技术在润滑油种类鉴别、润滑剂筛选、润滑性能评估和润滑监测等方面的研究进展,并对未来基于红外光谱技术应用于智能识别润滑油的研究进行了展望。

近红外光谱无损分析肉类品质的研究进展

近年来,近红外光谱分析技术以其无损、快速、高效、环境友好等特点,被广泛应用于肉类品质无损快速分析领域。然而,由于基质复杂、水分含量高,肉类常会对近红外光谱产生一定干扰,进而影响分析结果的准确度。为进一步明确近红外光谱分析技术在肉类品质无损分析方面的最新研究进展,本文对近年来近红外光谱技术在牛肉、羊肉、猪肉、鸡肉、水产品5种常见肉类的品质无损分析方面的应用进行梳理,包括品质检测、分类研究、真伪鉴别、质量安全4个方面,并就近红外光谱分析技术在肉类品质无损分析方面的应用进行总结与展望。可为近红外光谱分析技术在肉类品质无损检测、分类研究、真伪鉴别及质量安全方面的应用提供一定的参考与借鉴。

基于注意力机制残差神经网络的近红外芒果种类定性建模方法

现代光谱检测技术的飞速发展与深度学习紧密相关,作为一种端到端的模型,深度神经网络可以从光谱中得到更多信息,从而提升模型鲁棒性。为探究近红外光谱结合深度学习对芒果种类定性预测的可行性,提出一种基于卷积注意力机制(CBAM)的一维残差神经网络(1D-AD-ResNet-18)模型。为降低光谱中冗余信息的干扰,在传统一维残差神经网络(1D-ResNet-18)中嵌入CBAM卷积注意力模块,该模块可重点关注光谱局部有用信息;为避免梯度消失、过拟合情况发生,使用解决网络“退化”问题的ResNet-18。对于186个芒果样本,采用70%的样本进行训练,30%的样本进行测试,采用准确度(Accuracy)、精确率(Precision)、召回率(Recall)、F1值(F1-score)、宏观平均值(Macro-average)以及加权平均值(Weighted-average)作为模型评价指标。建立传统1D-ResNet-18、SNV-SVM和PCA-KNN三种对比模型,与上述三种方法作对比,所建立的1D-AD-ResNet-18模型取得最优预测结果,四种定性分析模型的准确率分别为96.42%,80.35%,76.78%和67.85%。结果表明,1D-AD-ResNet-18模型实现了对芒果种类的准确识别与分类,为近红外光谱定性分析芒果种类提供了新思路。

应用衰减全反射红外光谱无损鉴别中国传统手工纸的方法研究

纤维种类鉴别是古代纸张保护的重要基础工作。探究中国传统手工纸的无损纤维鉴别的方法对中国古籍、档案及纸质文物的研究和保护具有重要意义。该研究应用衰减全反射傅里叶变换红外光谱(ATRFTIR)对已知纤维成分的17类64种中国传统手工纸标准样品进行了无损分析,参考纤维素和木质素的红外峰位及部分纸样的X射线衍射(XRD)分析结果确认纸样中的所有红外谱峰归属;并采用分波段比较分析的方法对相似度极高的谱图进行比对分析,分别总结各类纸张在4000~1800、1800~1500、1500~1200和1200~600cm^(-1)四个波段的谱图特征;同时对4cm^(-1)精度的谱图进行二阶导数处理,分别总结各类纸张在1500~1200和1200~900cm^(-1)两个波段的二阶导数谱图特征;最后通过红外结晶指数及其他峰高和峰面积比值计算结果实现对纸类更细致的区分。应用上述方法对16个盲测样品进行了有效性测试,红外分析结果与显微纤维鉴别结果一致,初步证明了该方法在中国传统手工纸无损纤维鉴别上的可行性和有效性。实验结果表明,应用ATR-FTIR无损分析法可对麻、桑构皮、檀皮、瑞香皮和竹几大类原料的传统手工纸纤维类型作出快速而准确的鉴别,对于混料的宣纸也同样适用;但对于显微分析也难以鉴别的亲缘关系较近的植物原料如桑、构皮之间的精细鉴别仍有一定局限性。

基于近中红外光谱技术鉴别4种大黄鱼产地

目的基于近中红外光谱技术准确快速区分4种产地来源的大黄鱼(福建养殖大黄鱼、温州养殖大黄鱼、舟山养殖大黄鱼和舟山野生大黄鱼)。方法本研究应用傅里叶变换红外光谱仪对4种大黄鱼样本在4000~650 cm^(-1)的红外吸收指纹图谱进行测定,基于特征波段下的光谱吸收差异并结合主成分分析(principal component analysis,PCA)、聚类分析(cluster analysis,CA)和线性判别分析(linear discriminant analysis,LDA)、支持向量机(support vector machine,SVM)模型对大黄鱼样品进行产地区分。结果采用测定波段4000~650 cm^(-1)的全光谱采集信息经过Savitzky-Golay算法平滑预处理后建立的SVM模型效果最优,对4种大黄鱼样本的测试集准确率为83.3%。进一步对福建养殖大黄鱼和野生大黄鱼的产地区分方法优化后,选取特征波段3690~2800 cm^(-1)+1800~650 cm^(-1)的光谱信息经过一阶导数(first derivative,1stDer)、二阶导数(second derivative,2ndDer)和标准正态变换(standard normal variate transformation,SNV)3种方式预处理后建立LDA判别模型,光谱训练集与预测集的准确率均达到100%;3690~2800 cm^(-1)+1800~650 cm^(-1)的光谱信息经SNV、多元散射校正(multiplicative scatter correction,MSC)预处理后的PCA效果最佳,2种大黄鱼样本间彼此间距远、无重叠,且前两个主成分累计贡献率均在90%以上;经SNV预处理后的CA分析中,除21号野生大黄鱼外,其余产地相同的大黄鱼样本均各自聚为一类。结论基于近中红外光谱测定并结合化学计量学处理的方法能够对大黄鱼产地进行较准确地快速区分,从而为大黄鱼溯源鉴别提供技术支撑。

红外光谱的不同产地黑果腺肋花楸果实鉴别

黑果腺肋花楸是已被列入新食品原料名单中的小浆果,富含花青素等成分,在酒类、饮料、功能食品、化妆品等领域广泛应用,具有较高的经济价值。因受不同产地气候等环境因素及种植条件的影响,黑果腺肋花楸果实品质差异明显。为规范黑果腺肋花楸果品市场管理,以中红外光谱技术结合化学计量学方法对不同产地黑果腺肋花楸果实进行鉴别。采集15个产区共750份黑果腺肋花楸果实红外光谱数据,采用K-S样本划分法,按4∶1比例将样本划分为训练集和测试集,并进行多元散射校正(MSC)、标准正态化(SNV)、移动平滑(SG)、一阶导数(FD)、二阶导数(SD)等光谱预处理,与原始光谱进行支持向量机(SVM)建模识别效果对比,确定最佳光谱预处理方法,同时对最佳光谱数据进行归一化处理。采用竞争性自适应重加权算法(CARS)和连续投影算法(SPA)提取光谱特征信息,并结合随机森林(RF)、极限学习机(ELM)、支持向量机(SVM)、偏最小二乘-判别分析(PLA-DA)进行建模对比,确定最佳模型。结果表明,MSC为最佳光谱预处理方法,MSC-SVM训练集识别率为93.33%,测试集识别率为92.67%,能有效减少光谱采集时产生的随机误差。经CARS、SPA提取MSC特征光谱波长,进行4种算法的建模结果对比,确定SPA-SVM模型为最佳识别模型,其训练集与测试集识别率均为100%,且仅需16个波长点即可完成准确识别。红外光谱结合化学计量学方法,尤其是SPA-SVM模型,可准确鉴别黑果腺肋花楸果实产地,为黑果腺肋花楸果实产地溯源、质量评价提供快速、简便的方法支撑,为打造地区特色品牌提供技术基础。

基于傅立叶变换红外光谱分析的3种石首鱼科鱼类品种区分

为实现鲈形目石首鱼科中大黄鱼、小黄鱼和黄姑鱼的品种快速准确区分,应用傅立叶变换红外光谱仪对大黄鱼、小黄鱼和黄姑鱼肌肉样品在4000~650 cm-1范围内的近中红外光谱吸收特征进行了测定,基于特征波段下的光谱吸收差异并结合主成分分析(principal component analysis,PCA)、聚类分析(cluster analysis,CA)和线性判别分析(linear discriminant analysis,LDA)、支持向量机(support vector machine,SVM)数据建模对3种石首鱼类样品进行品种区分。结果表明:3种鱼类肌肉样本在特征波段1800~650 cm-1下的红外吸光谱选择一阶导数(first derivative,1st Der)+标准正态变换(standard normal variate transformation,SNV)和1st Der+多元散射校正(multiplicative scatter correction,MSC)2种方式预处理后进行PCA,3种鱼样本间彼此间距远、无重叠;特征波段1800~650 cm-1下的红外光谱数据经1st Der结合SNV预处理后进行聚类分析时各自聚为一类;特征波段1800~650 cm^(-1)下的红外光谱经过1st Der、SNV、MSC预处理建立的LDA模型和经过SNV、MSC预处理建立的SVM模型测试集准确率均为100%。研究结果表明利用近中红外光谱吸收特征差异,结合化学计量学分析手段能够有效对大黄鱼、小黄鱼和黄姑鱼进行品种判别,为水产品真实性鉴别提供技术参考。

基于深度残差网络和近红外光谱的煤矸石智能识别

该文开发了一种融合近红外光谱技术与一维残差深度网络(1D-ResNet)的煤炭及矸石快速分类方法。为保证实验样本的多样性,从河南、河北、山东3省份的多个煤矿中采集了430个煤炭与矸石样本,并基于欧氏距离对异常样本予以剔除,以获得高质量的建模数据集。在此基础上,为准确捕捉煤炭和矸石与其光谱特征之间的复杂映射关系,构建了基于1D-ResNet的分类模型,可在有效解决梯度消失问题的同时深度挖掘煤炭与矸石的光谱特性,获得高精度的分析结果。五折交叉验证结果显示,该模型的平均准确率达96.26%,显著优于支持向量机和随机森林等传统机器学习算法。在训练集和测试集上,该模型的损失函数变化趋势表现出较高的一致性,说明模型具备良好的泛化能力。测试发现,模型处理每一百个样本的推理时间仅为16.230 ms,进一步突显了其在煤炭与矸石在线分选领域的优势和潜在应用价值。

傅里叶变换红外光谱指纹图谱鉴别艾绒等级

提出了傅里叶变换红外光谱指纹图谱鉴别艾绒等级的方法,通过8种光谱预处理方法(去噪处理、高斯滤波、多元散射校正、标准正态变换、一阶导数+Savitzky-Golay(SG)平滑、二阶导数+SG平滑、一阶导数+Norris Gap、二阶导数+Norris Gap)和5种模式识别方法[反向传播神经网络(BP-NN)算法、遗传优化支持向量机(SVM-ga)、粒子群优化支持向量机(SVM-pso)、随机森林(RF)算法、K-最近邻(KNN)算法]的结合对比,得到鉴别艾绒等级的最佳模型。结果表明,艾绒的指纹图谱中有11个共有峰,对其进行主成分分析,得到9个主成分,累计方差贡献率达到99.67%。标准正态变换结合SVM-pso算法的鉴别效果最好,其训练集的鉴别正确率为100%,测试集的鉴别正确率为93.3%。

矿物药龙骨与伪品现代动物骨骼的对比研究

目的:从微观、图谱、化学成分上对龙骨进行真伪鉴定,为龙骨的质量控制和评价方法提供参考。方法:根据龙骨矿物特性,利用显微镜观察、近红外光谱、X射线荧光光谱仪进行鉴定研究,对比真伪品龙骨单偏光、正交偏光镜下特征,以正品龙骨的近红外光谱为参照,建立真伪龙骨定性鉴别模型,分析龙骨内元素检测结果。结果:正品龙骨呈磷灰石光学特性,与动物骨骼伪品龙骨有较大区别,正品较现代动物骨骼伪品P、Ca相对较少,但富集元素Sr、F,利用近红外特征谱段可建立有较好预测能力的相关系数模型。结论:偏光显微镜、近红外光谱法、X射线荧光光谱法可提高龙骨的真伪鉴别结果正确率。

基于波峰波谷特征提取技术的檀香紫檀光谱识别

伴随着经济水平的飞速增长,人们对红木产品的需求日益增大,而檀香紫檀作为一种珍贵红木因其在精神层面上符合人们的追捧且价格昂贵,具有巨大的商业价值而被不法分子替代和假冒。为维护木材市场的秩序和消费者的利益、实现对檀香紫檀的快速无损检测以及鉴别,有必要建立一种快速、可靠的檀香紫檀木材智能识别方法。采用近红外光谱(NIR)分析技术提取檀香紫檀及其相似血檀的光谱信息,利用定性分析方法、偏最小二乘判别分析(PLS-DA)和误差反向传播人工神经网络(BPNN)对光谱信息建立校正模型,进而对檀香紫檀及其相似木材血檀进行识别;通过分析比较这三种模型对这两类木材识别的优缺点和识别准确率,验证该方法对檀香紫檀和血檀识别的可行性。研究发现三种判别模型均能够快速识别木材光谱图像,具有对檀香紫檀和血檀进行快速无损分类识别的能力,在选取不同图像处理方法的情况下,三种判别模型显示的结果各不相同,识别结果也存在差异。进一步分析表明,在使用BPNN模型对光谱数据进行建模时,通过预处理将原始光谱数据波长范围866~2533 nm的波峰波谷特征值作为输入,当输入层节点数为24个特征值,隐含层13个神经元时,模型的均方根误差最小,准确率达到96.43%。此外,缩短光谱范围并不会提高模型的识别率,而在三种模型中,全波段范围的BPNN模型具有最高的识别率。实验结果表明,基于人工神经网络模型结合NIR特征提取技术识别檀香紫檀木材有着较高的识别准确率,其效果相较于前人有一定的提升。该研究有利于减少人工识别的主观性,使用计算机更能够缩短识别时间,较高的准确率可以帮助维护木材市场的秩序和消费者权益。同时,这也为实现檀香紫檀智能视觉识别提供了参考,为红木产业的可持续发展提供技术支持。