Data fusion of near-infrared and mid-infrared spectroscopy for rapid origin identification and quality evaluation of Lonicerae japonicae flos.

A data fusion strategy based on near-infrared (NIR) and mid-infrared (MIR) spectroscopy techniques were developed for rapid origin identification and quality evaluation of Lonicerae japonicae flos (LJF). A high-level data fusion for origin identification was formed using the soft voting method. This data fusion model achieved accuracy, log-loss value and Kappa value of 95.5%, 0.347 and 0.910 on the prediction set. The spectral data were converted to liquid chromatography data using a data fusion model constructed by the weighted average algorithm. The Euclidean distance and adjusted cosine similarity were used to evaluate the similarity between the converted and the real chromatographic data, with results of 247.990 and 0.996, respectively. The data fusion models all performed better than the models constructed using single data. This indicates that multispectral data fusion techniques have a wide range of application prospects and practical value in the quality control of natural products such as LJF.

An online automatic sorting system for defective Ginseng Radix et Rhizoma Rubra using deep learning.

Objective: To establish a deep-learning architecture based on faster region-based convolutional neural networks (Faster R-CNN) algorithm for detection and sorting of red ginseng (Ginseng Radix et Rhizoma Rubra) with internal defects automatically on an online X-ray machine vision system. Methods: A Faster R-CNN based classifier was trained with around 20 000 samples with mean average precision value (mAP) of 0.95. A traditional image processing method based on feedforward neural network (FNN) obtained a bad performance with the accuracy, recall and specificity of 69.0%, 68.0%, and 70.0%, respectively. Therefore, the Faster R-CNN model was saved to evaluate the model performance on the defective red ginseng online sorting system. Results: An independent set of 2 000 red ginsengs were used to validate the performance of the Faster R-CNN based online sorting system in three parallel tests, achieving accuracy of 95.8%, 95.2% and 96.2%, respectively. Conclusion: The overall results indicated that the proposed Faster R-CNN based classification model has great potential for non-destructive detection of red ginseng with internal defects.

Deep Learning Network of Amomum villosum Quality Classification and Origin Identification Based on X-ray Technology.

A machine vision system based on a convolutional neural network (CNN) was proposed to sort Amomum villosum using X-ray non-destructive testing technology in this study. The Amomum villosum fruit network (AFNet) algorithm was developed to identify the internal structure for quality classification and origin identification in this manuscript. This network model is composed of experimental features of Amomum villosum. In this study, we adopted a binary classification method twice consecutive to identify the origin and quality of Amomum villosum. The results show that the accuracy, precision, and specificity of the AFNet for quality classification were 96.33%, 96.27%, and 100.0%, respectively, achieving higher accuracy than traditional CNN under the condition of faster operation speed. In addition, the model can also achieve an accuracy of 90.60% for the identification of places of origin. The accuracy of multi-category classification performed later with the consistent network structure is lower than that of the cascaded CNNs solution. With this intelligent feature recognition model, the internal structure information of Amomum villosum can be determined based on X-ray technology. Its application will play a positive role to improve industrial production efficiency.

Simultaneous quantitative analysis of Escherichia coli, Staphylococcus aureus and Salmonella typhimurium using surface-enhanced Raman spectroscopy coupled with partial least squares regression and artificial neural networks.

Simultaneous detection of mixed bacteria accurately and sensitively is a major challenge in microbial quality control field. In this study, we proposed a label-free SERS technique coupled with partial least squares regression (PLSR) and artificial neural networks (ANNs) for quantitative analysis of Escherichia coli, Staphylococcus aureus and Salmonella typhimurium simultaneously. SERS-active and reproducible Raman spectra can be acquired directly upon the bacteria and Au@Ag@SiO2 nanoparticle composites on the surface of gold foil substrates. After applying different preprocessing models, SERS-PLSR and SERS-ANNs quantitative analysis models were developed to map SERS spectra of concentrations of the Escherichia coli, Staphylococcus aureus and Salmonella typhimurium, respectively. Both models achieved high prediction accuracy and low prediction error, while the performance of SERS-ANNs model in both quality of fit (R2 > 0.95) and accuracy of predictions (RMSE < 0.06) was superior to SERS-PLSR model. Therefore, it is feasible to develop simultaneous quantitative analysis of mixed pathogenic bacteria by proposed SERS methodology.

Laser-induced breakdown spectroscopy (LIBS) for the detection of exogenous contamination of metal elements in lily bulbs.

Natural products with the underground edible part have the risk of excessive heavy metals due to the influence of the growing environment. In this study, the content of five metal elements in lily bulbs was detected by laser-induced breakdown spectroscopy (LIBS). In view of the mutual interference among elements, multivariable analysis models were established to effectively eliminate the interference. The partial least squares regression (PLSR) multivariate analysis model was evaluated by combining different data preprocessing with variable selection methods to achieve the best fit. The results show that the best regression model for Cu, Pb, Zn, Al, and Mg content achieved the coefficients determination of prediction (Rp2) values of 0.9920, 0.9737, 0.9835, 0.9723 and 0.9939, respectively, and root mean square error of prediction (RMSEP) values of 3.2386 mg/kg, 5.8559 mg/kg, 4.6334 mg/kg, 6.0073 mg/kg and 2.8103 mg/kg, respectively. Comprehensively comparing the accuracy, robustness, and number of variables of each model, it can be found that the PLSR model on the least absolute shrinkage and selection operator (LASSO) achieved good results in the quantitative prediction model of three kinds of metal elements. This indicates the superiority of the LASSO-PLSR algorithm framework and confirms the feasibility of LIBS technology for the detection of various metal elements in natural products.

X-ray-based machine vision technique for detection of internal defects of sterculia seeds.

An online machine learning system based on X-ray nondestructive quality evaluation technique was developed to detect internal defects of boat-fruited sterculia seed. The X-ray images of boat-fruited sterculia seed were first acquired by the detection system. Then, a boat-fruited sterculia seed net (BSSNet) was trained to identify the defective boat-fruited sterculia seeds based on the X-ray images. The BSSNet was evaluated with the accuracy, precision, specificity, and sensitivity as 94.64%, 93.51%, 92.37%, and 96.64%, respectively. Further, three classical CNN models including VGG16, Resnet, and Inception were trained on the same dataset with accuracy of 95.71%, 94.29%, and 94.64%, respectively. Compared with the classical CNN models, the BSSNet achieved similar or higher accuracy in X-ray images classification. Finally, an independent dataset containing 200 X-ray images was used to validate the performance of the BSSNet and obtained an accuracy of 96.5%. The results presented above demonstrated that this classification method has a great potential for industrial applications. PRACTICAL APPLICATION: An X-ray online detection system integrated with a machine vision model was used to evaluate the quality of boat-fruited sterculia seed. A low-power x-ray detection system can detect internal defects of the object and ensure safety in the production process. The developed machine vision can sort the boat-fruited sterculia seed with an accuracy of 96.5%. The proposed nondestructive detection system showed a good potential to be used in industrial applications.

A feasibility research on the application of machine vision technology in appearance quality inspection of Xuesaitong dropping pills.

Defect detection is a critical issue for the quality control of dropping pills, which is a special dosage form of traditional Chinese Medicine. Machine vision is a non-destructing testing technology and cost-effective with high accuracy that can be used to predict the detects of both interior and exterior of the sample by employing the camera. In this research, a machine vision system for inspecting quality of the Xuesaitong dropping pills (XDPs) that include non-spherical, abnormal sizes and colors was developed to evaluate the appearance quality of XDPs rapidly and accurately. Firstly, 270 images of XDPs containing qualified and three different types of defects were collected. Subsequently, the processing of the XDPs images were carried out. Finally, Three defecting categories classification models were developed and compared based on contour and color features. The experimental results showed that the Random Forest outperformed all the explored models and the classification accuracy for non-spherical, abnormal sizes and colors reached 98.52%, 100.00% and 100.00%, respectively. In summary, the method established in this research is scientific, reliable, fast and accurate, which has great application potential and can provide technical support for the automatic defect detection of dropping pills.

Perceptual-Aware Sketch Simplification Based on Integrated VGG Layers.

Deep learning has been recently demonstrated as an effective tool for raster-based sketch simplification. Nevertheless, it remains challenging to simplify extremely rough sketches. We found that a simplification network trained with a simple loss, such as pixel loss or discriminator loss, may fail to retain the semantically meaningful details when simplifying a very sketchy and complicated drawing. In this paper, we show that, with a well-designed multi-layer perceptual loss, we are able to obtain aesthetic and neat simplification results preserving semantically important global structures as well as fine details without blurriness and excessive emphasis on local structures. To do so, we design a multi-layer discriminator by fusing all VGG feature layers to differentiate sketches and clean lines. The weights used in layer fusing are automatically learned via an intelligent adjustment mechanism. Furthermore, to evaluate our method, we compare our method to state-of-the-art methods through multiple experiments, including visual comparison and intensive user study.

A Strategy for Selecting "Q-Markers" of Chinese Medical Preparation via Components Transfer Process Analysis with Application to the Quality Control of Shengmai Injection.

Chinese medical preparation has complicated chemical constituents. Consequently, the proper quality control methods for these Chinese medical preparations have been great challenges to the traditional Chinese medicine modernization and internationalization. What components should be chosen for quality control is a big challenge in the development of traditional Chinese medicine. A new concept of "Quality Marker" was proposed by Liu et al. to solve this problem and established a new research paradigm for traditional Chinese medicine quality study. Several strategies were proposed by the researchers in traditional Chinese medicine, here, we used Shengmai injection as an example to discuss a strategy for selecting "Quality Markers" of Chinese medical preparation by the components transfer process analysis in the Shengmai injection manufacturing process. Firstly, a total of 87 compounds were identified or partially characterized in shengmai injection. Secondly, referenced to the quality control method in China pharmacopeia and considered the biomarkers in the original medicines and representative components in the manufacturing process, four ginsenosides in Panax ginseng (Hongshen), two compounds in Schisandra chinensis (Wuweizi), and a sugar from Ophiopogon japonicas (Maidong) were quantified. As a result, these seven representative compounds exhibited an acceptable transitivity throughout the Shengmai injection manufacturing process. Finally, combined with the active ingredients, components transfer process analysis, and comprehensive evaluation by "Spider-web" analysis, six compounds were selected as the quality markers for the quality control of Shengmai injection. Through this strategy of optimization for quality markers of Shengmai injection, we found that these six compounds could represent the main bioactive substances and be easily detected in the whole process of production. Furthermore, the quality control method was developed for quality assessment and control of these six quality markers in the Shengmai injection. The total content range of the selected quality markers in the 10 batches of the Shengmai injection is 13.844-22.557 mg/mL.