基于模糊模式的蝗虫图像识别方法

蝗灾是中国农业病虫害防治的重点,采用低空机载蝗虫预警系统技术,对拍摄图像中蝗虫识别和计数,实现蝗虫监测,可以弥补蝗虫地面人工监测的不足。以广东省清远市英德区域农田为试验区,采用数码相机采集蝗虫图像,对蝗虫区域和背景的RGB分量平均值进行对比分析,选用超G绝对值法进行灰度转换,实现蝗虫与背景分离。通过面积统计对比,确定单个蝗虫的面积和周长特征,建立单个蝗虫模糊集和粘连重叠蝗虫区域模糊集,采用最大隶属度原则可以判定蝗虫连通区域为单个蝗虫或是存在图像粘连重叠。用模糊识别方法对单个和粘连重叠的区域分别计算数量,准     

Use of differential scanning calorimetry (DSC) as a new technique for detection of adulteration in honeys. 1. Study of adulteration effect on honey thermal behavior.

Differential scanning calorimetry (DSC) was used to study the thermal behavior of authentic honeys (Lavandula, Robinia, and Fir honeys) and industrial sugar syrups. Thermal or thermochemical parameters such as the glass transition temperature (Tg), enthalpies of fusion (DeltaH(fus)), and heat capacity variation (DeltaC(p)) were measured. The syrups and honeys showed significant differences in thermal phenomena, as well as in their amplitude and position on the temperature scale. Results showed good reproducibility of the method for all samples studied. The effect of adulteration of honey with different amounts of syrup (5, 10, 20, 40, and 60%) was investigated. A linear relationship was found between the percentage of added syrup and the glass transition temperature. A similar relationship was obtained from the enthalpy of fusion results in the temperature range of 40-90 degrees C. Under applied conditions, the effects of adulteration of honeys by industrial syrups appeared to be detectable from a level as low as 5%.     

Technological classification of basque cider apple cultivars according to their polyphenolic profiles by pattern recognition analysis

The polyphenolic compositions of 31 Basque cider apple cultivars were determined in pulp, peel, and juice by high-performance liquid chromatography--diode array detection analysis of crude extracts and after thiolysis. Data sets, consisting of individual polyphenol concentrations, total procyanidin content, and the average degree of polymerization of procyanidins, were evaluated by multivariate chemometric techniques, to develop decision rules for classifying apple cultivars technologically into bitter and nonbitter categories. A preliminary study of the data structure was performed by cluster analysis and principal component analysis in each apple material. Bitter apple varieties presented higher contents of flavan-3-ols and/or dihydrochalcones than nonbitter cultivars. Different classification systems for the two categories on the basis of the chemical data were obtained applying several supervised pattern recognition procedures, such as linear discriminant analysis, K-nearest neighbors, soft independent modeling of class analogy, partial least-squares, and multilayer feed forward artificial neural networks. Excellent performance in terms of recognition and prediction abilities for both categories (100% of hits) was achieved in every case (pulp, peel, or juice). Polyphenolic profiles of apple pulp, peel, or juice provide enough information to develop classification criteria for establishing the technological group of apple cultivars (bitter or nonbitter).     

Determination of free proline and monosaccharides in wine samples by high-performance anion-exchange chromatography with pulsed amperometric detection (HPAEC-PAD)

A sensitive and selective analytical method for the simultaneous separation and quantitative determination of proline and free monosaccharides in wine samples by high-performance anion-exchange chromatography coupled with pulsed amperometric detection is described. Under optimized experimental conditions, a complete separation was obtained in less than 30 min, using an isocratic elution with 10 mM NaOH and 1 mM Ba(OAc)(2). No postcolumn addition of strong bases to the eluent for enhancing detection sensitivity was needed. Upon 25-fold sample dilution and purification to avoid interference of tannins, pigments, and phenolic compounds, the fingerprinting of common monosaccharides (i.e., arabinose, glucose, fructose, galactose, and xylose) and proline in wines, musts, and vinegars can be easily accomplished. The method allows high recovery and satisfies the necessary requirements for accuracy, repeatability, and sensitivity. Values obtained for proline content ranged from 470 to 1190 mg/L in "Aglianico" red wines (mean value, 870 +/- 192 mg/L, n = 21) and from 168 to 286 mg/L in white wines (mean value, 208 +/- 32 mg/L, n = 11). Lower levels were found in musts of red and white grapes, 550 and 87 mg/L, respectively. The lowest content of proline, ca. 10 mg/L, was found both in white and red vinegars.     

基于生物散斑图像和惯性矩谱分析的牛肉掺腐检测

牛肉掺假严重危害消费者的健康与经济利益,因此对牛肉掺假进行无损检测具有重要意义。该文基于生物散斑技术对牛肉掺假进行定量检测。试验将新鲜牛肉和非新鲜牛肉按不同比例(0、1%、3%、5%~60%(5%梯度)和100%)混合制备掺假样本,并采集样本的生物散斑图像。针对单列惯性矩(inertia moment,IM)表征样本生物活性存在稳定性差的问题,首次提出惯性矩谱(IM谱)分析的方法并用于建立基于支持向量回归机(support vector regression machine,SVR)的牛肉掺假检测模型。结果表明基于IM谱建立的SVR模型能较为准确预测牛肉中掺假物含量,校正集和测试集的决定系数分别为0.85和0.81,均方根误差分别为0.12和0.11。该研究证明了利用生物散斑技术和惯性矩谱分析方法对新鲜牛肉中掺杂腐败牛肉进行定量检测是可行的。     

1H NMR metabolite fingerprint and pattern recognition of mullet (Mugil cephalus) bottarga

Nuclear magnetic resonance (NMR) spectroscopy combined with multivariate data analysis (MVA) was used to investigate the molecular components of the aqueous extract of samples of bottarga, that is, salted and dried mullet (Mugil cephalus) roe, manufactured in Sardinia (Italy) from mullets of known and unknown geographical provenience. Principal component analysis (PCA) applied to the processed (1)H NMR spectra indicated that samples tend to cluster according to their geographical origin and also on the basis of storage and manufacturing procedures. The most important metabolites that characterized grouping of samples are the free amino acids methionine (Met), glutamate (Glu), histidine (His), phenylalanine (Phe), tyrosine (Tyr), and isoleucine (Ile); trimethylamine (TMA) and dimethylamine (DMA), both biomarkers of degradation; nucleotides and derivatives; choline (Cho) and phosphorylcholine (P-cho); and lactate (Lac).     

Classification of edible oils by employing 31P and 1H NMR spectroscopy in combination with multivariate statistical analysis. A proposal for the detection of seed oil adulteration in virgin olive oils

A combination of (1)H NMR and (31)P NMR spectroscopy and multivariate statistical analysis was used to classify 192 samples from 13 types of vegetable oils, namely, hazelnut, sunflower, corn, soybean, sesame, walnut, rapeseed, almond, palm, groundnut, safflower, coconut, and virgin olive oils from various regions of Greece. 1,2-Diglycerides, 1,3-diglycerides, the ratio of 1,2-diglycerides to total diglycerides, acidity, iodine value, and fatty acid composition determined upon analysis of the respective (1)H NMR and (31)P NMR spectra were selected as variables to establish a classification/prediction model by employing discriminant analysis. This model, obtained from the training set of 128 samples, resulted in a significant discrimination among the different classes of oils, whereas 100% of correct validated assignments for 64 samples were obtained. Different artificial mixtures of olive-hazelnut, olive-corn, olive-sunflower, and olive-soybean oils were prepared and analyzed by (1)H NMR and (31)P NMR spectroscopy. Subsequent discriminant analysis of the data allowed detection of adulteration as low as 5% w/w, provided that fresh virgin olive oil samples were used, as reflected by their high 1,2-diglycerides to total diglycerides ratio (D > or = 0.90).     

倒置残差网络结合注意力机制的掺假羊肉分类检测系统构建

针对羊肉精和染色剂作用下的猪肉掺假羊肉分类检测问题,该研究提出并建立了一种检测速度较快、精度较高的注意力机制结合倒置残差网络模型,同时基于智能手机开发了对应的快速、准确检测应用软件。首先,对羊肉、不同部位猪肉、不同掺假比例下的猪肉掺假羊肉的原始手机图像,使用数据增强方式进行数据扩充;其次,用倒置残差结构替换残差网络框架中的原有残差结构,以减少网络参数量并加快模型收敛速度,同时,引入注意力机制（Convolutional Block Attention Module,CBAM）,利用空间和通道特征对特征权重再分配,以强化掺假羊肉和羊肉之间的特征差异;然后,利用提出的注意力机制结合倒置残差网络（CBAM-Invert-ResNet）对样本进行训练并确定模型参数;最后,将训练好的网络模型移植到智能手机,以实现掺假羊肉的移动端检测。研究结果表明：与ResNet50和CBAM-ResNet50相比,Invert-ResNet50、CBAM-Invert-ResNet50模型的参数量分别减少了58.25%和61.64%,模型大小分别减小了58.43%和61.59%;针对背脊、前腿、后腿和混合部位数...     

Determination and differentiation of Flos Chrysanthemum based on characteristic electrochemical profiles by capillary electrophoresis with electrochemical detection

A high-performance capillary electrophoresis with electrochemical detection (CE-ED) method has been developed for the analysis of bioactive ingredients in Flos Chrysanthemum in this work. The effects of several factors such as the acidity and concentration of running buffer, the separation voltage, the applied potential, and the injection time were investigated. Under the optimum conditions, the eight analytes could be well separated within 20 min at the separation voltage of 14 kV in a 50 mmol/L Borax running buffer (pH 9.2). A 300 microm diameter carbon disk electrode has a good response at a potential of +950 mV (vs SCE) for all analytes. Good linear relationship was established over 3 orders of magnitude with detection limits (S/N = 3) that ranged from 1.9 x 10(-7) to 3.0 x 10(-8) g/mL. This proposed method has been successfully applied for the determination and differentiation of six kinds of popular Flos Chrysanthemum samples based on their characteristic electrochemical profiles, and the results are satisfactory.     

基于机器视觉的非结构化道路检测与障碍物识别方法

为了实现非结构化道路检测与障碍物的识别,提出了一种基于最小错误率贝叶斯决策与Hough变换相结合的非结构化道路检测与障碍物识别算法。算法首先将Otsu多阈值理论引入到最小错误率贝叶斯决策中并进行图像分割,然后利用Hough变换进行道路检测、提取出纯路面区域并再次进行路面分割,最后根据分割结果进行路面障碍物定位。结果表明,该算法能够有效实现非结构化道路的检测与障碍物的识别,在光影、照度变化、水渍等不利因素影响较小的情况下,具有较好的鲁棒性。     

基于级联卷积神经网络的番茄花期识别检测方法

对作物花期状态的准确识别是温室作物授粉的前提。为提高花期识别的准确率,该研究以温室番茄为例,提出一种基于级联卷积神经网络的番茄花期识别检测方法。首先采用改进的基于特征金字塔花束提取神经网络(Flower Extraction Feature Pyramid Networks, FE-FPN)实现番茄花束的局部区域提取,并采用Prim最小生成树对提取的花束区域图像进行识别优先级排序,然后按序将其输入到改进的Yolov3网络,实现番茄花朵不同花期的精准辨识检测。在包含4类花期、共1 600幅样本的番茄花束图像数据集上进行试验验证,本文方法对番茄不同花期的检测性能较好,平均检测精度达到了82.79%,平均单张检测时间为12.54 ms,各花期检测精度为花蕾期85.71%、全开期95.46%、谢花期62.66%、初果期88.34%;相比Mask R-CNN和空间金字塔池化网络（Spatial Pyramid Pooling Networks, SPP-Net）,平均检测精度提高了3.67和2.39个百分点,而且识别错误率比基础Yolov3网络降低了1.25个百分点。最后,将本文所提方法部署到大型玻璃温室环境下番茄授粉机器人上进行实际验证,识别准确率为76.67%,除去漏提取花束准确率达85.18%。研究结果可为设施番茄授粉机器人的精准作业提供重要依据。     

Effect of fertilizers on galanthamine and metabolite profiles in Narcissus bulbs by 1H NMR

Narcissus bulbs contain the biologically active alkaloid galanthamine, and Narcissus is being developed as a natural source of the molecule for the pharmaceutical industry. The effect of fertilizer on galanthamine production was investigated in a field study using a (1)H nuclear magnetic resonance (NMR) metabolite profiling approach. Galanthamine was quantitated and major metabolites in the bulbs were identified. The application of standard fertilization levels of nitrogen and potassium caused a significant increase in galanthamine as compared to a control. Multivariate data analysis of the (1)H NMR data revealed that applying double the standard level of nitrogen fertilizer resulted in production of more amino acids and citric acid cycle intermediates, but not more galanthamine. The results indicated that standard levels of fertilizer currently applied in The Netherlands are sufficient for optimal galanthamine accumulation in the bulbs. This study shows how (1)H NMR-based metabolic profiling can provide insight into the response of plant metabolism to agricultural practices.     

Phenotypic diversity and morphological characterization of Amygdalus L. species in Lebanon

In this paper, we report on morphological diversity of Amygdalus L. species native to Lebanon, following a countrywide survey of almond germplasm whereby a total of 149 accessions were collected throughout the country and characterized by thirteen quantitative and four qualitative traits. The results indicated that the genetic diversity of Amygdalus communis L., Amygdalus korshinskyi Hand.-Mazz., and Amygdalus orientalis Duh. in Lebanon is high. Principal component analysis revealed that nut weight, nut volume, nut width, kernel volume and shell strength had highest loading in the first component that accounted for 38.7% and 46.7% of total variation in A. communis and A. orientalis, respectively. In contrast, leaf traits were present in the second component which accounted for 18% and 23.2% of total variation in each species, respectively. No significant correlations were detected between leaf parameters and fruit traits in both species. The results indicated that quantitative leaf characters for all three species were determined by rainfall and not altitude whereby adjacent accessions located in drier areas had smaller leaf sizes than those located in more humid regions. Quantitative fruit characters did not seem to vary accordingly. Qualitative leaf traits in all three species reflected a variability which was independent of rainfall. A. communis populations showed high variability, suggesting that they could be a valuable source in almond improvement programs.     

Classification of German white wines with certified brand of origin by multielement quantitation and pattern recognition techniques

A procedure is proposed for the determination of the authenticity of white wines from four German wine-growing regions (Baden, Rheingau, Rheinhessen, and Pfalz) based on their content of some major, trace, and ultratrace elements. One hundred and twenty-seven white wine samples possessing a certificate of origin, all of the 2000 vintage, were analyzed. The concentrations of 13 elements (Li, B, Mg, Ca, V, Mn, Co, Fe, Zn, Rb, Sr, Cs, and Pb) were determined in wine diluted 1:20 by sector field inductively coupled plasma mass spectrometry (SF-ICP-MS). Indium was routinely used as internal standard. Supervised pattern recognition techniques such as discriminant analysis and classification trees were applied for the interpretation of the data. A quadratic discriminant analysis (QDA) allowed the four regions to be discriminated with 83% accuracy when using only eight variables (Li, B, Mg, Fe, Zn, Sr, Cs, and Pb), and the prediction ability for classifying new samples was 76%. By use of a second method, a decision tree, the classification of samples coming from the four regions could be performed with an accuracy of 84% when only four elements were used: Li (very low in samples from Baden), Zn (abnormally low in the samples from the Rheingau), and Mg and Sr (both important for the differentiation between Pfalz and Rheinhessen samples). For this method, the prediction ability was only 74% in the identification of unknown samples. The robustness of the QDA model was not good enough, and therefore the tree is better recommended for the classification of new wine samples from these areas of German wine production.     

基于奇异值分解和支持向量机的滚动轴承故障模式识别

提出了基于奇异值分解和支持向量机进行滚动轴承故障诊断的新方法。对故障轴承的状态特征提取和故障特征准确分类是解决该问题的两个关键。奇异值分解可以将高维相关变量压缩为低维独立的主特征矢量,而支持向量机可以完成模式识别和非线性回归。利用上述原理根据轴承振动信号的变化特征,采用奇异值分解对其提取状态主特征矢量,然后利用建立的支持向量机多故障分类器完成滚动轴承故障模式的识别。试验结果表明,奇异值分解后的主特征矢量与支持向量机相结合可以很好的分辨出轴承的正常和故障状态,并且对未知故障有良好的识别能力。与常用的人工神经网络方法相比,该诊断方法具有更好的有效性、鲁棒性和精确性。     

Arsenic species: effects on and accumulation by tomato plants.

The uptake of arsenic (As) species by Lycopersicum esculentum, growing under soilless culture conditions, was studied. A 4 x 3 x 2 factorial experiment was conducted with four As species (arsenite, arsenate, methylarsonate, and dimethylarsinate), three As concentrations (1, 2, and 5 mg L(-)(1)) and two tomato cultivars (Marmande and Muchamiel). The phytoavailability and phytotoxicity were primarily determined by the As species. The concentrations of As in plant increased significantly with increasing As concentration in solution. Both MA and DMA showed a higher upward translocation than arsenite and arsenate, and treatments with MA and DMA clearly reduced plant growth and fruit yield. The As concentration in tomatoes treated with arsenite or arsenate were within the range considered normal in food crops; however, the As concentration in tomatoes treated with MA and DMA were close to or even above the maximum limit. When tomato plants are exposed to high concentrations of As in nutrient solutions, they may uptake As to concentrations unacceptable for human food.     

基于卷积神经网络面部图像识别的拖拉机驾驶员疲劳检测

针对疲劳驾驶极易造成拖拉机交通事故这一问题,该文提出了一种基于卷积神经网络面部特征识别的拖拉机驾驶员疲劳检测方法。首先,利用伽马亮度校正对驾驶员面部图像进行光照预处理,再通过小波包去除图像中的椒盐噪声和高斯噪声,对预处理后的图像分别通过PCA-SCM人脸特征识别定位算法和基于人脸核心特征库及肤色模型的人脸识别算法进行驾驶员面部的识别定位,并通过比对这2种算法识别的偏差大小校验算法识别的有效性,以减小拖拉机工作振动时采样对图像中人脸定位精度的影响。将提取到的驾驶员面部图像输入到卷积神经网络进行深度学习和训练,并建立驾驶员疲劳视觉检测模型,从而实现基于拖拉机驾驶员面部图像的疲劳检测。统计训练过程中各项参数变化情况并进行T-SNE降维迭代分析,与其他常规方法相比,CNN在检测准确度和检测效率方面都有较为明显的优势。试验表明,所提出的检测模型准确率98.9%,图片识别效率38 ms/帧(Inter i7-4510U双核处理器),能够实现拖拉机驾驶员疲劳状况的实时检测,该研究可为解决疲劳驾驶这一安全问题提供参考。     

Development of a monoclonal antibody-based cELISA for the analysis of sulfadimethoxine. 1. Development and characterization of monoclonal antibodies and molecular modeling studies of antibody recognition

Sulfonamide antibiotics are used to treat a variety of bacterial and protozoan infections in cattle, swine, and poultry. Current residue methods for the analysis of sulfonamides in animal-based food products include bioassays, chromatographic methods (HPLC, GLC), and immunoassays. Most immunoassays have employed highly specific polyclonal antibodies. In this paper, we describe the isolation of monoclonal antibodies against sulfadimethoxine (SDM) that vary in their sensitivities and cross-reactivities against a large number of sulfonamides. The most sensitive monoclonal antibody, designated SDM-18, exhibits an IC(50) value for SDM of 1.53 ppb. Another monoclonal antibody, designated SDM-44, exhibits IC(50) values for six sulfonamides well below the established threshold level of 100 ppb for animal tissues. Molecular modeling studies of the cross-reactive drugs suggest that, depending on the monoclonal antibody, both steric and electronic features govern antibody binding. Due to the diversity of these monoclonal antibodies, it should be possible to design both compound- and class-specific monoclonal antibody-based immunoassays.     

Glutathione reductase in wheat grain. 1. Isolation and characterization

Durum wheat (Triticum durum, Desf.) endosperm of mature kernels contained a single form of glutathione reductase (GR); it appeared about the 18th day after anthesis while another isoform, present at the early stages of grain development, disappeared between the 20th and 30th days after flowering. The form that was present at grain maturity was isolated and characterized. It was composed of two monomers, each one having an apparent molecular mass of about 60 kDa. The K(m) values for NADPH and for GSSG were 3.7 and 9.1 microM, respectively, and the V(m) values for NADPH and for GSSG were 594 and 575 microkat.mg(-)(1) protein, respectively. The pH(i) of the enzyme was situated between pH 4.4 and 4.5. At a constant temperature of 25 degrees C, the optimum GR activity was found to be between pH 7.5 and 8.0. It was relatively resistant to high temperatures and was very resistant to very low temperatures.