基于可见/近红外高光谱成像技术的鸡蛋新鲜度无损检测

以400~1 000nm高光谱系统获得鸡蛋样本的高光谱图像,利用蒙特卡洛法检测异常样本,采用不同预处理方法处理原始光谱;应用竞争性正自适应加权算法(Competitive Adaptive Reweighted Sampling,CARS)、遗传偏最小二乘法(Genetic Algorithms PLS,GAPLS)和间隔蛙跳法(Interval Random Frog,IRF)对预处理后光谱数据提取特征波长;分别建立基于全光谱和特征波长的偏最小二乘回归(Partial Least Squares Regression,PLSR)和最小二乘支持向量机(Least Squares Support Vector Machine,LS-SVM)鸡蛋新鲜度预测模型。结果表明:标准正态变量变换(Standardized Normal Variate,SNV)法为最优预处理方法;利用CARS、GAPLS和IRF分别选出8,35,74个特征波长;基于GAPLS提取的特征波长的LS-SVM模型最优,其校正相关系数(Rc)为0.899,预测相关系数(Rp)为0.832。表明基于高光谱成像技术的鸡蛋新鲜度无损检测是可行的。     

Detection of insect-damaged wheat kernels using near-infrared hyperspectral imaging

Insect damage in wheat adversely affects its quality and is considered one of the most important degrading factors in Canada. The potential of near-infrared (NIR) hyperspectral imaging for the detection of insect-damaged wheat kernels was investigated. Healthy wheat kernels and wheat kernels visibly damaged by Sitophilus oryzae, Rhyzopertha dominica, Cryptolestes ferrugineus, and Tribolium castaneum were scanned in the 1000–1600 nm wavelength range using an NIR hyperspectral imaging system. Dimensionality of the acquired hyperspectral data was reduced using multivariate image analysis. Six statistical image features (maximum, minimum, mean, median, standard deviation, and variance) and 10 histogram features were extracted from images at 1101.69 and 1305.05 nm and given as input to statistical discriminant classifiers (linear, quadratic, and Mahalanobis) for classification. Linear discriminant analysis and quadratic discriminant analysis classifiers correctly classified 85–100% healthy and insect-damaged wheat kernels.     

Feasibility of Vis/NIR spectroscopy for detection of flaws in hazelnut kernels

The feasibility of Vis/NIR spectroscopy for detection of flaws in hazelnut kernels (Corylus avellana L. cv. Tonda Gentile Romana) is demonstrated. Feature datasets comprising raw absorbance values, raw absorbance ratios (Abs[λ₁]:Abs[λ₂]) and differences (Abs[λ₁] − Abs[λ₂]) for all possible pairs of wavelengths from 306.5 nm to 1710.9 nm were extracted from the spectra for use in an iterative LDA routine. For each dataset, several spectral pretreatments were tested. Each group of features selected was subjected to Partial Least Squares Discriminant Analysis (PLS-DA), Receiver Operating Characteristics (ROCs) analysis, and evaluation of performance through the Area Under ROC Curve. The best result (5.4% false negative, 5.0% false positive, 5.2% total error) was obtained using a Savitzky–Golay second derivative on the dataset of raw absorbance differences. The optimal features were Abs[564 nm]–Abs[600 nm], Abs[1223 nm]–Abs[1338 nm] and Abs[1283 nm]–Abs[1338 nm]. The results indicate the feasibility of a rapid, online detection system.     

Detection of fusarium damage in Canadian wheat using visible/near-infrared hyperspectral imaging

Fusarium damage in wheat may reduce the quality and safety of food and feed products. In this study, the use of hyperspectral imaging was investigated to detect fusarium damaged kernels (FDK) in Canadian wheat samples. More than 5,200 kernels, representing seven major Canadian wheat classes, with varying degree of infection symptoms ranging from sound through mild to severe were imaged in the visible-NIR (400–1,000 nm) wavelength range. Partial least squares discriminant analysis (PLS-DA) was used to segregate kernels into sound and damaged categories based on kernel mean spectra. A universal PLS-DA model based on four wavelengths was able to detect FDK in all seven classes with an overall accuracy of 90 % and false positives of 9 %.     

NIR高光谱成像技术检测冷鲜羊肉嫩度

利用900¹700 nm近红外高光谱成像系统对冷鲜羊肉嫩度进行快速无损检测研究。采集冷鲜羊肉（1⁸ d）表面的高光谱散射图像,提取样本感兴趣区域反射光谱曲线并用剪切力值表征冷鲜羊肉的标准嫩度。以原始光谱、特征区域光谱和Savitzky-Golay卷积平滑预处理光谱建立冷鲜羊肉嫩度的偏最小二乘回归（PLSR）模型,预处理的特征区域光谱建立的模型效果更优。结果表明:特征区域光谱可有效替代全波段光谱,经过S-G卷积平滑预处理后,模型预测效果最佳,预测相关系数（Rp）和均方根误差（RMSEP）分别为0.773和1.060。研究表明:利用近红外高光谱成像技术结合偏最小二乘回归法对冷鲜羊肉嫩度的快速无损检测是可行的。      

Non-destructive assessment of instrumental and sensory tenderness of lamb meat using NIR hyperspectral imaging

The purpose of this study was to develop and test a hyperspectral imaging system (900–1700 nm) to predict instrumental and sensory tenderness of lamb meat. Warner–Bratzler shear force (WBSF) values and sensory scores by trained panellists were collected as the indicator of instrumental and sensory tenderness, respectively. Partial least squares regression models were developed for predicting instrumental and sensory tenderness with reasonable accuracy (R_cv = 0.84 for WBSF and 0.69 for sensory tenderness). Overall, the results confirmed that the spectral data could become an interesting screening tool to quickly categorise lamb steaks in good (i.e. tender) and bad (i.e. tough) based on WBSF values and sensory scores with overall accuracy of about 94.51% and 91%, respectively. Successive projections algorithm (SPA) was used to select the most important wavelengths for WBSF prediction. Additionally, textural features from Gray Level Co-occurrence Matrix (GLCM) were extracted to determine the correlation between textural features and WBSF values.     

Rapid screening of DON contamination in whole wheat meals by Vis/NIR spectroscopy and computer vision coupling technology

This work intends to develop an online experimental system for screening of deoxynivalenol (DON) contamination in whole wheat meals by visible/near-infrared (Vis/NIR) spectroscopy and computer vision coupling technology. Spectral and image information of samples with various DON levels was collected at speed of 0.15 m s⁻¹ on a conveyor belt. The two-type data were then integrated and subjected to chemometric analysis. Discriminant analysis showed that samples could be classified by setting 1000 μg kg⁻¹ as the cut-off value. The best correct classified rate obtained in prediction was 93.55% based on fusion of spectral and image features, with reduced prediction uncertainty as compared to single feature. However, quantification of DON by quantitative analysis was not successful due to poor model performance. These results indicate that, although not accurate enough to provide conclusive result, this coupling technology could be adopted for rapid screening of DON contamination in cereals and feeds during processing.     

便携式近红外水果糖度分析模型简化研究

为了实现便携式近红外光谱仪水果糖度现场快速分析,将桃、梨和苹果的光谱进行二阶导数和卷积平滑处理后,利用组合移动窗口偏最小二乘法选择信息变量建立PLS模型,利用遗传偏最小二乘法选择信息变量建立MLR模型.分析表明,桃、梨和苹果PLS模型的RMSEP分别为0.417、0.372和0.654,其RSDP分别为4.685%、3.348%和4.111%;MLR模型的RMSEP分别为0.381、0.382和0.550,其RSDP分别为4.281%、3.438%和3.457%,模型预测精度均满足现场检测应用要求.结果表明, 用SCMWPLS和GA-PLS可以提取最有效信息变量,模型更加简洁、数据运算量也更少,模型适用于便携式近红外光谱仪器.     

Thermal imaging to detect infestation by Cryptolestes ferrugineus inside wheat kernels

Canada's zero tolerance for live insects in grain received from farmers, and shipped to domestic and export buyers, has necessitated the development of an accurate insect detection method. An infrared thermal imaging system was developed to detect infestation by six developmental stages (four larval instars, pupae and adults) of Cryptolestes ferrugineus under the seed coat on the germ of the wheat kernels. The artificially infested wheat kernels were removed from the incubation room (30 °C), refrigerated (5 °C) for 60 s, maintained at ambient conditions for 20 s, and imaged using a thermal camera to identify each developmental stage (n=283). The means of the highest 5% and 10% of all temperature values on the surface of the grain were significantly higher (=0.05) for grains having young larvae inside and lower for grains having pupae inside. Temperature distribution on the surface of the infested kernels with different stages of C. ferrugineus was highly correlated with the respiration rate of each developmental stage (r=0.83–0.91). The overall classification accuracy for a quadratic function was 83.5% and 77.7% for infested and sound kernels, respectively, and for a linear function, it was 77.6% and 83.0% for infested and sound kernels, respectively, in pairwise discriminations. Thermal imaging has the potential to identify whether the grain is infested or not, but is less effective in identifying which developmental stage is present.     

Non-destructive determination of water-holding capacity in fresh beef by using NIR hyperspectral imaging

This study was carried out for post-mortem non-destructive prediction of water holding capacity (WHC) in fresh beef using near infrared (NIR) hyperspectral imaging. Hyperspectral images were acquired for different beef samples originated from different breeds and different muscles and their spectral signatures were extracted. Both principal component analysis (PCA) and partial least squares regression (PLSR) models were developed to obtain an overview of the systematic spectral variations and to correlate spectral data of beef samples to its real WHC estimated by drip loss method. Partial least squares modeling resulted in a coefficient of determination (R_CV²) of 0.89 and standard error estimated by cross validation (SECV) of 0.26%. The PLSR loadings showed that there are some important absorption peaks throughout the whole spectral range that had the greatest influence on the predictive models. Six wavelengths (940, 997, 1144, 1214, 1342, and 1443 nm) were then chosen as important wavelengths to build a new PLS prediction model. The new model led to a coefficient of determination (R_CV²) of 0.87 and standard error estimated by cross validation (SECV) of 0.28%. Image processing algorithm was then developed to transfer the predicting model to each pixel in the image for visualizing drip loss in all portions of the sample. The results showed that hyperspectral imaging has the potential to predict drip loss non-destructively in a reasonable accuracy and the results could be visualised for identification and classification of beef muscles in a simple way. In addition to realize the difference in WHC within one sample, it was possible to accentuate the difference in samples having different drip loss values.     

Prediction of beef quality attributes using VIS/NIR hyperspectral scattering imaging technique

Hyperspectral imaging images were used to predict fresh beef tenderness (WBSF: Warner-Bratzler Shear Force) and color parameters (L^∗, a^∗, b^∗). Sixty-five fresh strip loin cuts were collected from 33 carcass after 2 days postmortem. After acquiring hyperspectral images, the samples were vacuum packaged and aged for 7 days, and then the color parameters and WBSF of the samples were measured as references. The optical scattering profiles were extracted from the images and fitted to the Lorentzian distribution (LD) function with three parameters. LD parameters, such as the scattering asymptotic vale, the peak height, and full scattering width were determined at each wavelength. Stepwise discrimination was used to identify optimal wavelengths. The LD parameters’ combinations with optimal wavelengths were used to establish multi-linear regression (MLR) models to predict the beef attributes. The models were able to predict beef WBSF with R_cv = 0.91, and color parameters (L^∗, a^∗, b^∗) with R_cv of 0.96, 0.96 and 0.97, respectively.     

基于可见/近红外光谱和变量优选的南水梨糖度在线检测

利用可见/近红外光谱技术对南水梨糖度进行在线检测研究。南水梨样本以0.3m/s速度传输,并采用USB4000光谱仪在470~1 150nm波段范围内采集南水梨样本的光谱。然后,利用3种变量选择方法对波长变量进行筛选,应用偏最小二乘(PLS)方法分别建立南水梨糖度的在线预测模型,并分析预测模型性能的优劣。结果表明:可见/近红外光谱技术结合变量选择方法在线检测南水梨的糖度是可行的;竞争自适应重加权采样(CARS)方法优于无信息变量消除(UVE)及连续投影算法(SPA);CARS方法可以有效简化预测模型并提高预测模型的性能;南水梨全光谱PLS及CARS—PLS糖度预测模型的预测集相关系数和预测均方根误差(RMSEP)分别为0.940,0.951和0.467%,0.420%。     

Scanning electron microscopy of Fusarium damaged kernels of spring wheat

Kernels of five wheat cultivars (Triticum aestivum) of different bread-making quality were examined. Grown under field conditions, heads of wheat were inoculated in the flowering stage with an aqueous suspension of Fusarium culmorum conidia. Wheat heads were collected from the control and inoculated plots at full maturity. Control (non-inoculated) kernels without any symptoms of disease and Fusarium damaged kernels (FDK) were examined under scanning electron microscopy (SEM). Examination of the FDK fraction confirmed localisation of Fusarium hyphae on the surface and inside the tissues of kernels. Observations of the endosperm from Fusarium infected kernels revealed presence of fungal hyphae in the endosperm and some characteristic structural changes in many of its regions, such as partial or complete lack of the protein matrix, damage to large and small starch granules caused by fungal amylolytic enzymes, disappearance of small starch granules as the colonisation progressed, complete disappearance of the starchy endosperm under severe infection. Fungal colonisation of the endosperm and structural changes in its area were highly variable traits within the FDK fraction of a given cultivar.     

Application of NIR hyperspectral imaging for discrimination of lamb muscles

The potential of near-infrared (NIR) hyperspectral imaging system coupled with multivariate analysis was evaluated for discriminating three types of lamb muscles. Samples from semitendinosus (ST), Longissimus dorsi (LD) and Psoas Major (PM) of Charollais breed were imaged by a pushbroom hyperspectral imaging system with a spectral range of 900-1700 nm. Principal component analysis (PCA) was used for dimensionality reduction, wavelength selection and visualizing hyperspectral data. Six optimal wavelengths (934, 974, 1074, 1141, 1211 and 1308 nm) were selected from the eigenvector plot of PCA and then used for discrimination purpose. The results showed that it was possible to discriminate lamb muscles with overall accuracy of 100% using NIR hyperspectral reflectance spectra. An image processing algorithm was also developed for visualizing classification results in a pixel-wise scale with a high overall accuracy.     

Application of Vis/NIR Spectroscopy for Chinese Liquor Discrimination

As one of the most widely consumed alcoholic beverages, Chinese liquor varies greatly in price, flavor, and quality. This diversity calls for effective and reliable discrimination methods. In an attempt to find the best liquor discrimination method, this study used different methods to analyze and identify 730 Chinese liquor samples including 22 kinds, ten brands, and six flavors. These samples, covering most of the famous liquors in China, were analyzed by visible and near-infrared (Vis/NIR) spectroscopy and modeled by three classification methods including supporting vector machine, soft independent modeling of class analogy, and linear discriminate analysis based on principal component analysis (PCA-LDA). Pretreatments and parameters for each model were optimized, and models discrimination ability was compared. The research finds that PCA-LDA was the best model with an average prediction rate of 98.94 % in the training set and 95.70 % in the test set. The correct rates for brands, flavor styles, ages, and alcohol degrees were all higher than 95 %. It shows that Vis/NIR is a reliable, inexpensive, and effective tool for Chinese liquors discrimination.     

Fast Discrimination of Apple Varieties Using Vis/NIR Spectroscopy

We evaluated the potential of visible/near-infrared (Vis/NIR) spectroscopy for its ability to nondestructively differentiate apple varieties. The apple varieties used in this research included, Fuji apples, Red Delicious apples, and Copefrut Royal Gala apples. The chemometrics procedures applied to the Vis/NIR data were principal component analysis (PCA), wavelet transform (WT), and artificial neural network (ANN). The apple varieties could be qualitatively discriminated in the PC1-PC2 space resulted from PCA. Wavelet transform was used as a tool for dimension reduction and noise removal, reducing spectral to wavelet components. Wavelet components were utilized as input for three-layer back propagation ANN model. WT-ANN model gave the highest level of correct classification (100%) of the apple varieties.     

Olive oil nutritional labeling by using Vis/NIR spectroscopy and compositional statistical methods

Food nutritional labeling is compulsory in the European Union since 13 December 2016. The olive oil fatty acid composition shows high variation depending mainly on the variety. Thus, olive oil nutritional labeling is problematic for the industry. Besides, the analysis of all batches of olive oil using the official methods is expensive. Therefore, the olive oil industry is seriously concerned about solutions for nutritional labeling. In this study, a new rapid technique to measure the nutrients for the olive oil nutritional labeling, is assessed. A novel partial least squares (PLS) calibration model using log-ratio coordinates has been formulated and successfully tested for predicting the percentages of monounsaturated, saturated, and polyunsaturated fatty acids based on visible and near infrared spectroscopy. The model provided accuracy suitable for labeling, under the rules in force in the European Union. The error was generally much lower than the tolerance.Industrial relevanceThe approach here proposed can be a suitable solution for olive oil nutritional labeling, which is a current challenge for the olive oil industry.     

On-line and non-destructive measurement of core temperature in heat treated fish cakes by NIR hyperspectral imaging

During industrial heat treatment of food products the core temperature is a critical control parameter with respect to food quality and in particular food safety. This paper presents a method based on near-infrared (NIR) spectroscopy that enables on-line and non-contact monitoring of the complete product volume on a typical industrial belt cooking system. Two NIR systems (760–1040 nm) were evaluated on heat treated fish cakes, one point measurement system and one hyperspectral imaging system. Both systems measured several millimetres into the product. Core temperature in the fish cakes (at 10 mm depth) varied between 53 and 99 °C. The point system performed best with a root mean square error of prediction of 2.3 °C, while the imaging system was less accurate with an error of 4.5 °C. It was demonstrated that temperature changes down to 11–13 mm depth in the fish cakes could be registered by the NIR point system.Industrial relevanceDuring industrial heat treatment of food products the core temperature is a critical control parameter with respect to food quality and in particular food safety. Especially for ready-made products this is important since they can be consumed without further heat treatment. Today, most temperature measurements during processing are typically based on spot checks on a small number of products. The core temperature of heat treated products is usually the most critical and needs to be measured by insertion of thermo couplers. This procedure is insufficient since it leaves the producer with a large degree of uncertainty; only a few products are checked and a very tiny volume of the checked products is actually measured. Due to these limitations, current practise is to over-cook much of the food to ensure that everything has reached the critical core temperature. This might reduce quality of the end product and also requires overspending of energy. In the food industry there is a need for non-contact on-line temperature measurements for improved control of the cooking process. The ideal system should be able to log the temperature in the entire production volume.The method presented in this article can allow complete monitoring of the heat treated products. In this way the producer could have full control of the heating process and be sure that sufficient core temperature is reached in all product units. Such a system can also be used to control the temperature to a certain target that ensures safe products of high quality.