基于傅里叶近红外光谱和电子鼻技术的苹果霉心病无损检测

为研究傅里叶近红外光谱技术(Fourier transform near infrared spectroscopy,FT-NIRS)和电子鼻技术分别结合化学计量学方法对苹果霉心病的判别效果,以“红富士”霉心病苹果和健康苹果为试材,利用近红外光谱技术,基于主成分分析建立Fisher判别和多层感知器(multi-layer perceptron,MLP)神经网络模型;同时利用电子鼻技术分别结合Fisher判别、MLP神经网络和径向基函数神经网络3种化学计量学的方法建立判别模型。根据建模集和验证集的预测准确率综合考虑,基于主成分分析建立的MLP神经网络模型和电子鼻结合MLP神经网络模型对苹果霉心病的判别效果最好,验证集中的正确判别率分别达到87.7%和86.2%。说明电子鼻和近红外光谱技术均可以较好地判别苹果霉心病。     

基于深度信念网络的近红外光谱鉴别莲子粉掺假

为快速鉴别莲子粉真伪,利用近红外光谱技术对莲子粉掺杂进行鉴别。基于已知利用支持向量机(support vectormachine,SVM)对光谱数据分类的结果,在未知样品类别的情况下使用基于深度信念网络(deepbelief network,DBN)进行判别。结果表明,当训练集数目达到600时,SVM模型对掺入不同比例各类其他作物粉的平均识别率达到98%;基于DBN模型能够有效识别掺杂了各类其他作物粉的莲子粉,极个别掺杂比例的平均识别率在96%左右。采用DBN算法避免了当前深层神经网络易陷入局部最优和无大量标签样本的情况。近红外光谱技术结合DBN为农产制品掺假的快速检测提供了新的尝试。     

基于透射光谱的苹果霉心病多因子无损检测

针对目前苹果霉心病难以检测的问题,提出一种基于透射光谱的苹果霉心病多因子无损检测方法,通过融合多波段透射光谱与苹果直径,构建苹果霉心病判别模型,实现了苹果霉心病无损检测。搭建光谱测试范围在200～1 025 nm的透射光谱采集平台,实验获取232 个苹果样本的透射光谱数据,采用游标卡尺获得苹果直径数据。采用杂散光校正,非线性校正对苹果透射原始光谱进行预处理,选取与霉心病发病相关的12 个波段透射光强值,结合苹果的直径进行主成分分析,将分析的结果作为自变量,建立苹果霉心病Fisher判别模型。经过异校验验证,模型总体识别率为93.1%,而仅采用透射光谱构建的模型识别率为91.37%。结果表明,基于透射光谱与直径结合的多因子检测方法可实现苹果霉心病的准确判定,为苹果霉心病的快速、无损检测提供可行思路。     

基于电子鼻表征霉心病苹果特征气味及无损检测模型建立

为探究电子鼻检测技术对霉心病苹果的判别效果,以富士健康苹果和霉心病苹果为试材,基于SIMCA软件对采集的不同病变程度霉心病苹果的电子鼻信息进行表征,基于SPSS 23.0软件建立霉心病苹果Fisher函数、多层感知器神经网络(muhilayer perceptron neural network, MLPNN)和径向基函数神经网络(radial basis function neural network, RBFNN)判别模型。结果表明,健康果的特征传感器有W5C、W1C和W3C,重度果的特征传感器有W1S、W2S、W1W和W5S;MLPNN模型的判别效果最好,其对训练集和验证集的总体预测判别率分别为88.61%和88.46%;RBFNN模型的判别效果次之,其对训练集和测试集的总体预测判别率分别为93.50%和80.95%;Fisher判别函数判别效果最差,其对训练集和验证集的总体预测判别率分别为91.50%和79.27%。另外,3种判别模型对健康果和重度果都有很好的判别效果,对健康果和轻度果的判别效果不理想,需要在今后的研究中进一步优化。     

基于FT-NIR和电子鼻的苹果水心病无损检测

探讨傅里叶变换近红外光谱技术和电子鼻技术应用于苹果水心病检测的可行性。以277?个“秦冠”水心病苹果和健康苹果为试材,分别采集每个样本在12?000～4?000?cm－1波数范围的近红外光谱和10?个传感器的电子鼻信号,用不同预处理的近红外光谱方法提取主成分建立Fisher判别模型;同时电子鼻结合3?种化学计量学的方法进行建模。结果表明,经一阶导数（9?点平滑）预处理的近红外光谱,提取前20?个主成分建立的Fisher判别模型效果最好,对未知样本的正确判别率达100%;电子鼻分别结合Fisher判别、多层感知器神经网络和径向基函数神经网络判别模型对未知样本的识别率为89.7%、89.5%和85.7%。故利用近红外光谱和电子鼻技术分别结合化学计量学的方法可快速、无损检测苹果的水心病。其中,近红外光谱技术结合Fisher判别对苹果水心病的识别率最高,是一种准确可靠的测定方法。     

基于高光谱成像技术预测牡蛎干制加工过程中的水分含量

提出一种应用高光谱成像技术结合化学计量学检测牡蛎干制加工过程中水分含量的方法。采用高光谱成像系统,在400～1 100 nm范围内,采集到5个干燥时期的100个牡蛎干样本高光谱图像。提取所有样本感兴趣区域的平均光谱数据,对原始光谱数据进行多元散射校正(MSC)、卷积平滑(S-G)预处理,采用相关系数法提取8个特征波长。基于所提取的特征波长,建立光谱数据与水分含量的多元线形回归(MLR)和BP神经网络模型。结果表明:两种模型均有较好的预测效果。MLR模型的校正集、预测集和交叉验证集的相关系数较BP神经网络低;校正集、预测集和交叉验证集均方根误差分析结果表明,BP神经网络效果较MLR好。高光谱成像技术结合化学计量学方法可检测牡蛎干制过程中水分含量的变化。     

基于电子鼻技术的马铃薯真菌性腐烂病早期检测

确定马铃薯致腐菌种,并建立快速检测方法对腐烂样本进行识别。本研究通过微生物及分子生物学方法对马铃薯致腐菌进行鉴定,采用电子鼻方法对样本进行检测,建立模型对不同感染阶段样本进行识别。结果显示:马铃薯主要致腐烂菌为出芽短梗霉,建立的K最近邻判别模型中,训练集与预测集识别率分别为90%和85%;建立的BP网络判别模型中,训练集与预测集判别率分别达到93.75%和90%,各腐烂阶段能够较好地被识别。研究结果为后期电子鼻技术应用至马铃薯腐烂病检测提供理论基础。     

基于高光谱反透射图像的新疆冰糖心红富士水心鉴别

采用高光谱成像技术结合化学计量法,采集新疆冰糖心红富士好果与水心病果样本在波长范围380～1 004 nm的可见近红外高光谱反透射图像,选取感兴趣区域获得平均光谱,对原始光谱采用直接差分一阶求导等9种光谱预处理方法,再分别用主成分分析、快速独立分量分析、相关系数法完成数据降维,结合贝叶斯判别、K最近邻法、马氏距离判别、最小二乘支持向量机、二次线性判别方法识别是否有水心病。结果表明,主成分分析提取前15主成分,采用标准正态变量变换-主成分分析-最小二乘支持向量机与多元散射校正-主成分分析-最小二乘支持向量机模型识别效果最优,校正集和预测集识别率分别为100%和91.2%。     

基于近红外光谱和共轭梯度神经网络的板栗褐变检测

为了实现板栗褐变的无损检测,本实验以“毛板红”板栗为样品,在12000 ～4000 cm-1范围内采集带壳和去壳板栗4个褐变等级的近红外光谱,用Savitzky-Golay平滑和标准正态变量变换(SNV)方法对光谱原始数据进行预处理,采用主成分分析法提取光谱的特征信息,建立基于共轭梯度调整算法的BP神经网络(SBP)识别板栗褐变模型.结果表明,对去壳板栗,最佳主成分因子数为8时,网络训练集和测试集对板栗褐变识别准确率最好,分别为100％和98.7％;对带壳板栗,最佳主成分因子数为10,网络训练集和测试集对板栗褐变识别准确率最好,分别为65.3％和64.4％.最后比较了所建网络与传统的基于梯度下降算法的BP神经网络(GBP)与径向基函数(RBF)网络的性能,验证集结果表明,构建的基于共轭梯度调整算法神经网络模型(SBP)效果好于GBP和RBF,对去壳板栗和带壳板栗褐变识别准确率分别为100％和66.7％.     

基于数据的废纸配比预测纸浆白度的研究

针对实际生产过程中根据人工经验配比废纸用量导致纸浆性能与预期差别较大的现状,本研究利用纸厂提供的废纸配比和纸浆性能检测数据,使用BP神经网络和支持向量机（SVM）的建模方法,分别采用全部样本数据和样本平均值数据建立基于废纸配比的纸浆白度预测模型。研究结果表明,在模型预测精度、预测稳定性以及模型训练时间等方面,以样本平均值数据作为建模数据集,使用SVM方法建立的纸浆白度预测模型,具有较好的预测精度(2.42%)和良好的稳定性(0.58%),且模型训练时间短(0.2 s),可以满足实际生产过程的需要。     

Surface Gloss Evaluation of Apples Based on Computer Vision and Support Vector Machine Method

This paper investigates computer vision applications for surface gloss evaluation to determine a quick surface gloss evaluation method for apples. “Red Fuji” apples were wax-coated with different concentrations of shellac solutions to obtain the apple samples with different levels of surface gloss. The surface gloss values and the color scales of the apple samples were detected using a pinhole gloss meter and a color meter. The apple sample images were captured and processed, and the color parameters of the high light areas were extracted. Support vector machine (SVM) regression and classification models were built to predict the surface gloss values and the surface gloss levels of apples, respectively. The results showed that to predict the surface gloss of apple samples, the correlation coefficients of the SVM regression model were 0.94 and 0.90 for the training and the testing groups, respectively. The classification accuracy rates of the SVM classification model for the training and the testing groups were 100 and 96.7%, respectively. Finally, apple surface gloss level classification software was developed, which showed good operating results for both classification accuracy rates and calculation speed. This paper provided a new surface gloss evaluation method based on computer vision for apples.     

基于机器视觉的苹果在线分级

通过CCD相机动态采集苹果两个面的实时图像,提出了泛洪填充+自适应Ostu阈值分割算法提取苹果的轮廓,采用最小外接圆法对苹果上表面图像进行处理得到苹果果径,采用最小外接矩形法对苹果侧表面图像进行处理提取苹果果形特征;将图像进行RGB到HSV空间转换,提取苹果的着色度、果锈,以及疤痕特征,采用基于改进粒子群算法的SVM决策树的分类方法进行苹果的分级。结果表明,该方法对特级果、一级果、二级果和等外果的识别准确率分别达96%,94%,98%,98%,分级速率达4个/s,可以满足苹果在线分级的要求。     

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.     

NIRS Detection of Moldy Core in Apples

Moldy core of apples is undetectable until the fruit is cut open or bitten into, therefore it can pose serious problems to both producer and consumer. Removal of diseased fruits prior to storage would be most desirable. The objective of this study was to evaluate the ability of VIS-NIR minispectrometers to detect moldy core in apples, on line. An apparatus which is qualified for on-line Near-Infrared Spectroscopy (NIRS) measurements was developed based on an off-the-shelf minispectrometer. Apples, cv. Red Delicious, were collected from several orchards before and during the commercial harvest, and were stored at 0°C pending the tests. The data were analyzed by chemometric procedures, specifically, by partial least squares regression (PLSR), and were classified by means of canonical discriminant analysis. The canonical variables were represented by the latent variables of PLS models based on the spectra. The accuracy of the classification results was high, in light of doubts regarding the moldy fraction threshold of 5%; in such a case the mold covers only the seed carpals of the fruit, where it might remain without really damaging the fruit. Improvements should aim to reduce errors in classifying low-level damage, and also in misclassifying some healthy fruits. The rate of testing (1 s per fruit) is acceptable for quality-control purposes, but should be accelerated for future packing-line implementation.     

Enhanced preservation effects of sugar apple fruits by salicylic acid treatment during post‐harvest storage

BACKGROUND: Salicylic acid (SA) is recognised as an endogenous signal, mediating in plant defense and against pathogens. It has been reported that SA treatment can reduce decay and extend storage life of various fruit, such as bananas, peaches and apples. RESULTS: Physiological and biochemical responses in harvested sugar apple fruit (Annona squamosa L.) to SA at 0.4, 0.8 and 1.2 mmol L^?1 were investigated during post‐harvest storage. Results indicated that SA treatments lowered respiration, increased activities of antioxidant enzymes of superoxide dismutase (SOD), peroxidase (POD), catalase (CAT) and ascorbate peroxidase (APX), decreased lipoxygenase (LOX) activity and correspondingly lowered Malondialdehyde (MDA) contents in treated fruits was observed as compared to the control. Moreover, production of superoxide free radical (O₂^{?, ?}) and ethylene was significantly decreased in the treated fruits. Total soluble solids, total soluble sugar, softness and decay rate were significantly lowered in treated fruits, and in turn a delay in the fruits ripening process was achieved after 10 days of storage. CONCLUSION: SA has positive effects in maintaining membrane integrity and in delaying fruit ripening process, which results in improved storability of sugar apple. Copyright © 2008 Society of Chemical Industry     

采用二次连续投影法和BP人工神经网络的寒富苹果病害高光谱图像无损检测

为提供苹果病害在线、快速、无损检测的理论依据,采用高光谱成像技术进行了北方大面积种植的寒富苹果病害无损检测研究。寒富苹果的主要病害有炭疽病、苦痘病、黑腐病和褐斑病害。为选择较少的有效波长而利于在线快速检测,首先采集高光谱苹果图像,分割出感兴趣区域并提取光谱信息,然后采用连续投影算法(successive projections algorithm,SPA)从全波长(500~970 nm)中提取了10个特征波长SPA1(502、573、589、655、681、727、867、904、942 nm和967 nm),再对这10个特征波长采用连续投影算法提取3个特征波长SPA2(681、867 nm和942 nm)。最后利用全波长光谱信息、SPA1提取的10个特征波长的光谱信息和SPA2提取的3个特征波长的光谱信息作为输入矢量采用线性判别分析、支持向量机和BP人工神经网络(BP artificial neural network,BPANN)模型进行苹果病害的检测。通过对检测结果分析,最终选择SPA2-BPANN为最佳检测方法,训练集检测率达100%,验证集检测率达100%。结果表明,高光谱成像技术可以有效对苹果病害进行检测,所获得的特征波长可为开发多光谱成像的苹果品质检测和分级系统提供参考。     

联合图像最优特征提取及改进RBF神经网络的苹果质量估计

目的：以阿克苏苹果为例,设计一种联合图像最优特征提取和改进RBF神经网络学习的苹果质量估计方法,以克服人工分级称重成本高、误差大的缺陷。方法：首先,建立苹果图像采集系统,得到苹果前景图像信息;其次,设计苹果图像特征集合最佳子集提取策略,将最佳子集提取过程转化为目标函数优化问题,并利用改进的离散蝗虫优化算法进行求解,从而得到最佳苹果图像特征子集;最后,构建基于RBF神经网络学习的苹果质量估计模型,将最佳特征子集作为网络输入,并采用蝗虫优化算法优化配置RBF神经网络超参数,从而实现对苹果质量的有效估计。结果：所提苹果质量估计方法精度更高,质量估计值平均相对误差率为1.23%。结论：该方法可以有效实现苹果质量预估,也能够推广应用到其他类似轴对称形状的水果质量估计。