Pork product adulteration detection method and system based on hyperspectral imaging

Abstract

The invention discloses a pork product adulteration detection method and system based on hyperspectral imaging. The method comprises the following steps: acquiring hyperspectral image information of pork products adulterated with different concentrations of kappa-carrageenan; the collected hyperspectral image is processed through related software; acquiring a hyperspectral image of the pork product to be detected and drawing a corresponding spectrogram; and analyzing based on the comparison data and the sample data. The system comprises a training set construction module, a network training module and an output module. According to the scheme, the requirement for sample preparation is not high, and the method has the characteristics of no damage, rapidness and high sensitivity. The method can be widely applied to the field of component detection.

Description

Technical Field

[0001] The present invention relates to the field of ingredient detection, and in particular to a method and system for detecting adulteration of pork products based on hyperspectral imaging. Background Art

[0002] Meat products are the main source of high-quality protein for people. Due to problems such as uneven color, high cooking loss, water loss during pressing, and poor texture and taste, water-soluble colloids are usually added to improve the quality of meat products [1]. However, the addition of water-soluble colloids will bring some potential problems to colloid-injected meat. Although the ions carried by these colloids have a positive effect on the quality of meat, injecting them into meat products will destroy the cell and tissue structure of the meat, allowing metal ions, microorganisms and viruses in the environment to enter the cell and tissue structure of the meat, which may cause serious heavy metal pollution and increase the risk of consumers ingesting pathogens, thereby causing foodborne diseases. Therefore, colloid-injected meat has potential food safety issues.

[0003] Currently, the commonly used method for detecting glue-injected meat is liquid chromatography, which is costly, complex and destructive. In addition, Raman spectroscopy cannot accurately identify the OH bonds that produce fluorescence, making this method difficult to be widely used. Summary of the invention

[0004] In view of this, in order to solve the problem that most of the existing meat product adulteration detection methods are destructive detection and difficult to be applied to on-site real-time detection of injected meat, the present invention proposes a pork product adulteration detection method based on hyperspectral imaging, which includes the following steps:

[0005] S1. Obtain hyperspectral image information of pork products adulterated with different concentrations of κ-carrageenan. Specifically, put pork products adulterated with different concentrations (0-10%) of κ-carrageenan on a sample tray, set multiple replicate groups for each concentration of injected meat sample, and then place them under the near-infrared light (900-1700nm) and visible light (400-1000nm) lenses of a hyperspectral instrument for scanning, obtain corresponding images through the data acquisition software Lumo Scanner (64-bit), and associate related data to build a comparison set;

[0006] S2. Process the collected hyperspectral images through relevant software to remove the interference of background such as sample boxes and sample plates on the data, generate the area of ​​interest and the reflectance value of each wavelength, and use the reflectance value of each wavelength to draw a spectrum to obtain comparative data;

[0007] S3, obtaining a hyperspectral image of the pork product to be tested and drawing a corresponding spectrum diagram to obtain sample data;

[0008] S4. Based on the comparison data and sample data, the characteristic peaks and troughs and their position shifts are observed to preview the differences. Then, the reflectance values ​​of each wavelength are used for partial least squares discriminant analysis (PLS-DA) and support vector machine (SVM) analysis. The near-infrared and visible light spectra, PLS-DA (full wavelength, characteristic wavelength), and SVM visual classification diagrams are output to quickly identify whether pork products are adulterated with κ-carrageenan.

[0009] The present invention also proposes a pork product adulteration detection system based on hyperspectral imaging, the system comprising:

[0010] The data set construction module obtains the hyperspectral image information of pork products adulterated with different concentrations of κ-carrageenan and constructs a comparison set;

[0011] The first drawing module processes the collected hyperspectral image through relevant software, draws a spectrum diagram, and obtains comparative data;

[0012] The second drawing module obtains the hyperspectral image of the pork product to be tested and draws the corresponding spectrum to obtain sample data;

[0013] The comparison and analysis module performs analysis based on the comparison data and sample data to obtain the test results.

[0014] Based on the above scheme, the present invention provides a pork product adulteration detection method and system based on hyperspectral imaging. The hyperspectral instrument is used to collect sample information, and the data is fitted and classified in combination with machine learning, so as to quickly and non-destructively identify the situation of pork product adulteration. It can not only effectively detect the situation of excessive addition of κ-carrageenan to pork products in the market, but also ensure the health rights and interests of consumers and promote the development of the meat products industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 This is a spectrum diagram of pork products adulterated with different concentrations of κ-carrageenan in Example 1 of the present invention;

[0016] Figure 2 This is a schematic diagram of the results of using PLS-DA to quickly detect κ-carrageenan injected into pork products based on full-band hyperspectral imaging technology in Example 2 of the present invention;

[0017] Figure 3 This is a schematic diagram of the result of visual classification based on characteristic wavelengths selected from the entire band in Example 3 of the present invention;

[0018] Figure 4It is a confusion matrix and performance diagram of Example 4 of the present invention using hyperspectral imaging technology based on SVM model analysis;

[0019] Figure 5 This is a schematic diagram of the results of the embodiment 2-1 of the present invention using the VIP scores of PLS-DA in full band based on hyperspectral imaging technology to quickly detect the injection of κ-carrageenan into pork products;

[0020] Figure 6 Embodiment 5 of the present invention is a confusion matrix and performance diagram based on KNN model analysis using hyperspectral imaging technology;

[0021] Figure 7 1 is a confusion matrix and performance diagram of Example 6 of the present invention using hyperspectral imaging technology based on NB model analysis;

[0022] Figure 8 7 is a confusion matrix and performance diagram of the XGB model analysis using hyperspectral imaging technology according to Example 7 of the present invention;

[0023] Fig. 9 1 is a confusion matrix and performance diagram of embodiment 8 of the present invention using hyperspectral imaging technology based on DT model analysis;

[0024] Fig.10 It is a confusion matrix and performance diagram of Example 9 of the present invention using hyperspectral imaging technology based on SVM model analysis. DETAILED DESCRIPTION

[0025] Hyperspectral imaging (HSI) is a type of spectroscopy technology, which is divided into near-infrared and visible light according to different wavelengths. HSI has high resolution, wide wavelength coverage, and the combination of image and spectral information, as well as a continuous spectral reflectance curve close to the object being detected, which can obtain accurate detection results. The HSI spectral absorption band is very sensitive to high-energy chemical bonds (mainly OH, CH) in organic matter, which are abundant in glue-injected meat, and HSI can accurately identify OH bonds that produce fluorescence, which is not available in Raman spectroscopy. In addition, HSI is a direct, non-destructive and environmentally friendly detection method that supports the simultaneous acquisition of multi-component spectral and spatial information, and has low sample requirements, making it suitable for on-site real-time detection of glue-injected meat.

[0026] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0027] It should be noted that, for the convenience of description, only the parts related to the relevant invention are shown in the drawings. In the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other.

[0028] It should be understood that the "system", "device", "unit" and / or "module" used in this application is a method for distinguishing different components, elements, parts, portions or assemblies at different levels. However, if other words can achieve the same purpose, the word can be replaced by other expressions.

[0029] As shown in this application and claims, unless the context clearly indicates an exception, the words "a", "an", "a kind" and / or "the" do not refer to the singular, but also include the plural. Generally speaking, the terms "include" and "comprise" only indicate the inclusion of clearly identified steps and elements, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements. The elements defined by the sentence "includes a..." do not exclude the existence of other identical elements in the process, method, commodity or device that includes the elements.

[0030] In the description of the embodiments of the present application, "plurality" means two or more than two. The following terms "first" and "second" are used for descriptive purposes only and are not to be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features.

[0031] In addition, flow charts are used in the present application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed accurately in order. On the contrary, the various steps may be processed in reverse order or simultaneously. At the same time, other operations may also be added to these processes, or a certain step or several steps of operations may be removed from these processes.

[0032] Example 1

[0033] A method for preliminary analysis of adulteration of pork products with κ-carrageenan based on hyperspectral imaging combined with spectral images.

[0034] (1) Data acquisition: Obtain hyperspectral image information of pork products adulterated with different concentrations of κ-carrageenan. Pork products adulterated with different concentrations (0-10%) of κ-carrageenan were placed on a sample tray, and 10 replicates were set for each concentration of injected meat sample. They were then placed under the near-infrared light (900-1700nm) and visible light (400-1000nm) lenses of a hyperspectral instrument for scanning, and the corresponding images were acquired using the data acquisition software Lumo Scanner (64-bit).

[0035] (2) Data processing:

[0036] a) Use ENVI 5.6 (64-bit) software to process the collected hyperspectral images, remove the interference of background such as sample boxes and sample plates on the data, and obtain the reflectance values ​​of each wavelength of near-infrared and visible light in the area of ​​interest;

[0037] b) Use the reflectance values ​​of each wavelength of near-infrared and visible light to draw spectra, observe the characteristic peaks and troughs and their position shifts, and preview the differences

[0038] (3) Data output: Output the near-infrared and visible light spectra analyzed in (2).

[0039] Reference Figure 1 (Left: near infrared; right: visible light), the peaks and troughs, reflectance values ​​and offset positions of the injected and non-injected meat were observed using the spectrum graph drawn in Example 1. In the near infrared band of Example 1, the injected and non-injected meat showed the same peaks and troughs, and the reflectance value of the injected meat was higher than that of the non-injected meat as a whole. In addition, compared with the non-injected meat, the injected meat with different concentrations of adulterated injected meat in the visible light band was shifted to the short-wave direction at wavelengths of 434nm, 550nm and 569nm, respectively, to 426nm, 545nm and 564nm; 489nm and 574nm were shifted to the long-wave direction, respectively, to 500nm and 580nm.

[0040] Example 2

[0041] A method for quickly identifying adulterated κ-carrageenan in pork products based on full-band hyperspectral imaging data combined with PLS-DA.

[0042] (1) Data acquisition: Based on the reflectance values ​​of each wavelength of near-infrared and visible light obtained in Example 1.

[0043] (2) Data processing: Near-infrared and visible light PLS-DA analysis was performed using the full-band online analysis website MetaboAnlyst (https: / / www.metaboanalyst.ca / )

[0044] (3) Data output: Output the near-infrared and visible light PLS-DA visualization 2D graphs analyzed in (2) to quickly identify adulterated κ-carrageenan in pork products.

[0045] The hyperspectral imaging in Example 1 was used to extract the reflectance values ​​of each wavelength of near-infrared and visible light for full-wavelength PLS-DA analysis, and the PLS-DA model was observed to perform 2D visual recognition and classification of glue-injected meat (2%-10%) and non-glue-injected meat (0%). Figure 2(Left: near infrared; right: visible light), in Example 2, the 2D visualization to identify whether pork products are mixed with κ-carrageenan shows that visible light is better than near infrared, and can better qualitatively detect whether pork products are injected with κ-carrageenan. Among them, component 1 of visible light (93.6%) is much larger than component 1 of near infrared (76.3%), and the 2D spatial distribution range occupied by injected meat is larger than that of non-injected meat. The results show that the full band of visible light based on hyperspectral imaging can better and quickly detect the addition of κ-carrageenan in pork products.

[0046] Example 3

[0047] A method for quickly identifying adulterated κ-carrageenan in pork products based on hyperspectral imaging characteristic band data combined with PLS-DA.

[0048] (1) Data acquisition: Use the characteristic wavelengths (959nm, 1133nm, 1136nm, 1140nm, 1143nm, 1147nm, 1150nm, 1154nm, 1157nm, 1161nm, 1164nm, 1168nm, 1171nm, 1175nm, 1178nm) selected from the near-infrared band within the range of 900-1200nm and the characteristic wavelengths (599nm, 596nm, 601nm, 604nm, 593nm, 561nm, 564nm, 558nm, 442nm, 566nm, 439nm, 607nm, 447nm, 556nm, 444nm) and their reflectance values ​​selected from the visible light band within the range of 400-615nm.

[0049] (2) Data processing: Near-infrared and visible light PLS-DA analysis was performed using the characteristic band online analysis website MetaboAnlyst (https: / / www.metaboanalyst.ca / )

[0050] (3) Data output: Output the near-infrared and visible light PLS-DA visualization 2D graphs analyzed in (2), which is a method for rapid quantitative identification of adulterated κ-carrageenan in pork products.

[0051] Using the full-band data obtained in Example 1, quantitative detection of glue-injected meat with different concentrations was performed according to the steps of Example 3. Figure 3 (Left: near infrared; right: visible light), in Example 3, the 2D classification effect of visible light is the best, and the 2D classification effect of near infrared is second. The results show that compared with the cumbersome liquid chromatography, the hyperspectral imaging technology has the advantages of non-destructive and rapid quantitative detection of pork products injected with different concentrations of κ-carrageenan, especially the 2D visualization effect diagram of PLS-DA based on visible light.

[0052] Example 4

[0053] A method for quickly identifying adulterated κ-carrageenan in pork products based on hyperspectral imaging characteristic band data combined with SVM model.

[0054] (1) Data acquisition: Example 4 of the present invention uses the reflectance values ​​of each wavelength of near-infrared and visible light obtained in Example 1.

[0055] (2) Data processing: The Python program was used to perform SVM analysis on the near-infrared and visible light data, with the key parameters set as random_state = 45 and test_size = 0.3 to obtain the confusion matrix diagram and the model analysis performance results: accuracy, precision, precision, and F1-score.

[0056] (3) Data output: Output the near-infrared and visible light SVM models analyzed in (2) to analyze the various result indicators and quickly and quantitatively identify pork products adulterated with κ-carrageenan.

[0057] Using the full-band data obtained in Example 1, quantitative detection of glue-injected meat with different concentrations was performed according to the steps of Example 4. Figure 4 In Example 4, the accuracy of near infrared quantitative detection of κ-carrageenan in pork reached 78%, precision 75%, recall rate 80%, F1-score 76%, and the classification model was good; the accuracy of visible light quantitative detection of κ-carrageenan in pork reached 94%, precision 96%, recall rate 94%, F1-score 94%, and the classification model was the best. The results show that the hyperspectral imaging technology combined with the SVM model has the advantages of non-destructive, rapid and accurate quantitative detection of κ-carrageenan injected into pork products with different concentrations, among which the results obtained by SVM model analysis using the full band of visible light are the best.

[0058] Based on the above scheme, the method model for detecting the illegal addition of κ-carrageenan in pork products provided by the present invention has the characteristics of wide application range, no use of organic reagents, safety and greenness; the method model for detecting the illegal addition of κ-carrageenan in pork products provided by the present invention has low requirements for sample preparation, and has the characteristics of non-destructive, rapid and high sensitivity.

[0059] Example 2-1

[0060] The characteristic wavelengths were selected in combination with the VIP scores of PLS-DA to provide a reference for the rapid identification of adulterated κ-carrageenan in pork products.

[0061] (1) Data acquisition: Example 2-1 of the present invention is based on the analysis of the reflectance values ​​of each wavelength of near-infrared and visible light of PLS-DA in Implementation Case 2.

[0062] (2) Data processing: The full-band online analysis website MetaboAnlyst (https: / / www.metaboanalyst.ca / ) was used to perform near-infrared and visible light PLS-DA analysis to obtain VIP scores.

[0063] (3) Data output: Output the VIP scores of near-infrared and visible light PLS-DA analyzed in (2), and obtain the wavelengths with VIPscores>1, providing a reference for PLS-DA to perform characteristic wavelength analysis.

[0064] Using the VIP scores obtained by PLS-DA analysis in Example 2-1, 15 near-infrared or visible light wavelengths and their reflectance values ​​with VIP scores greater than 1 are selected to provide a basis for secondary PLS-DA analysis. Figure 5 (Left figure: near infrared; right figure: visible light), in Example 2-1, the characteristic wavelengths in the range of 900-1200nm in the near infrared band are selected based on high VIP scores as 959nm, 1133nm, 1136nm, 1140nm, 1143nm, 1147nm, 1150nm, 1154nm, 1157nm, 1161nm, 1164nm, 1168nm, 1171nm, 1175nm, 1178nm, and the characteristic wavelengths in the range of 400-615nm in the visible light band are selected based on high VIP scores as 599nm, 596nm, 601nm, 604nm, 593nm, 561nm, 564nm, 558nm, 442nm, 566nm, 439nm, 607nm, 447nm, 556nm, 444nm. The results showed that high VIP scores could provide a reference for secondary PLS-DA analysis and provide a methodological reference for the rapid detection of κ-carrageenan added to pork products using hyperspectral imaging technology.

[0065] Example 5

[0066] A method for quickly identifying adulterated κ-carrageenan in pork products using the KNN model.

[0067] (1) Data acquisition: Example 5 of the present invention utilizes the reflectance values ​​of each wavelength of near-infrared and visible light obtained in Example 1.

[0068] (2) Data processing: The KNN (K-NearestNeighbour) analysis was performed on the near-infrared and visible light data using the Python program. The key parameters were set as random_state = 45 and test_size = 0.3 to obtain the confusion matrix diagram and the model analysis performance results: accuracy, precision, precision, and F1-score.

[0069] (3) Data output: Output the near infrared and visible light KNN models analyzed in (2) to analyze the various result indicators and quickly and quantitatively identify pork products adulterated with κ-carrageenan.

[0070] Using the full-band data obtained in Example 1, quantitative detection of glue-injected meat with different concentrations was performed according to the steps of Example 5. Figure 6 In Example 5, the accuracy of near-infrared quantitative detection of κ-carrageenan in pork reached 67%, precision 58%, recall rate 66%, F1-score 61%, and the classification model accuracy was low; the accuracy of visible light quantitative detection of κ-carrageenan in pork reached 78%, precision 82%, recall rate 78%, F1-score 76%, and the classification model accuracy was high. The results show that the visible light band based on hyperspectral imaging technology combined with the KNN model can non-destructively and quickly and accurately quantitatively detect pork products injected with different concentrations of κ-carrageenan.

[0071] Example 6

[0072] A method for quickly identifying adulterated κ-carrageenan in pork products based on hyperspectral imaging characteristic band data combined with NB model.

[0073] (1) Data acquisition: Example 6 of the present invention utilizes the reflectance values ​​of each wavelength of near-infrared and visible light obtained in Example 1.

[0074] (2) Data processing: Use Python program to perform NB( Bayes) analysis, with key parameters set as random_state=45 and test_size=0.3, to obtain the confusion matrix diagram and model analysis performance results: accuracy, precision, accuracy, and F1-score.

[0075] (3) Data output: Output the near-infrared and visible light NB model analysis results of (2) to quickly and quantitatively identify pork products adulterated with κ-carrageenan.

[0076] Using the full-band data obtained in Example 1, quantitative detection of glue-injected meat with different concentrations was performed according to the steps of Example 6. Figure 7In Example 6, the accuracy of near infrared quantitative detection of κ-carrageenan in pork reached 50%, precision 60%, recall rate 54%, F1-score 52%, and the accuracy of the classification model was only 50%, and the model was not applicable; the accuracy of visible light quantitative detection of κ-carrageenan in pork reached 72%, precision 81%, recall rate 69%, F1-score 69%, and the accuracy of the classification model was also low. The results show that the visible light band based on hyperspectral imaging technology combined with the NB model is not applicable for non-destructive, rapid and accurate quantitative detection of pork products injected with different concentrations of κ-carrageenan.

[0077] Example 7

[0078] A method for quickly identifying adulterated κ-carrageenan in pork products based on hyperspectral imaging characteristic band data combined with the XGB model.

[0079] (1) Data acquisition: Example 7 of the present invention utilizes the reflectance values ​​of each wavelength of near-infrared and visible light obtained in Example 1.

[0080] (2) Data processing: The Python program was used to perform XGB (eXtremeGradient Boosting) analysis on the near-infrared and visible light data, with the key parameters set as random_state = 45 and test_size = 0.3 to obtain the confusion matrix diagram and the model analysis performance results: accuracy, precision, precision, and F1-score.

[0081] (3) Data output: Output the near-infrared and visible light XGB model analysis results of (2) to quickly and quantitatively identify pork products adulterated with κ-carrageenan.

[0082] Using the full-band data obtained in Example 1, quantitative detection of glue-injected meat with different concentrations was performed according to the steps of Example 7. Figure 8 In Example 7, the accuracy of near infrared quantitative detection of κ-carrageenan in pork reached 44%, precision 47%, recall rate 43%, F1-score 43%, and the accuracy of the classification model was less than 50%, and the model was not applicable; the accuracy of visible light quantitative detection of κ-carrageenan in pork reached 89%, precision 93%, recall rate 86%, F1-score 87%, and the classification model accuracy was high. The results show that the visible light band based on hyperspectral imaging technology combined with the XGB model can be used to quickly and accurately quantitatively detect pork products injected with different concentrations of κ-carrageenan without loss.

[0083] Example 8

[0084] A method for quickly identifying adulterated κ-carrageenan in pork products based on hyperspectral imaging characteristic band data combined with DT model.

[0085] (1) Data acquisition: Example 8 of the present invention utilizes the reflectance values ​​of each wavelength of near-infrared and visible light obtained in Example 1.

[0086] (2) Data processing: The Python program was used to perform decision tree (DT) analysis on the near-infrared and visible light data, with the key parameters set as random_state = 45 and test_size = 0.3 to obtain the confusion matrix diagram and the model analysis performance results: accuracy, precision, accuracy, and F1-score.

[0087] (3) Data output: Output the near-infrared and visible light DT models analyzed in (2) to analyze the various result indicators and quickly and quantitatively identify pork products adulterated with κ-carrageenan.

[0088] Using the full-band data obtained in Example 1, quantitative detection of glue-injected meat with different concentrations was performed according to the steps of Example 8. Fig. 9 In Example 8, the accuracy of near infrared quantitative detection of κ-carrageenan in pork reached 56%, precision 71%, recall rate 59%, F1-score 56%, and the accuracy of the classification model was only 56%, with low model accuracy; the accuracy of visible light quantitative detection of κ-carrageenan in pork reached 72%, precision 60%, recall rate 67%, F1-score 62%, and the classification model accuracy was high. The results show that the visible light band based on hyperspectral imaging technology combined with the DT model can be used to quickly and accurately quantitatively detect pork products injected with different concentrations of κ-carrageenan, but the accuracy is low.

[0089] Example 9

[0090] A method for rapid qualitative analysis of adulterated κ-carrageenan in pork products based on hyperspectral imaging characteristic band data combined with SVM model.

[0091] (1) Data acquisition: Example 9 of the present invention utilizes the reflectance values ​​of each wavelength of near-infrared and visible light obtained in Example 1.

[0092] (2) Data processing: The Python program was used to perform SVM analysis on the near-infrared and visible light data, with the key parameters set as random_state = 35 and test_size = 0.3 to obtain the confusion matrix diagram and the model analysis performance results: accuracy, precision, precision, and F1-score.

[0093] (3) Data output: Output the near-infrared and visible light SVM models analyzed in (2) to analyze the various result indicators and quickly and qualitatively identify pork products adulterated with κ-carrageenan.

[0094] Using the full-band data obtained in Example 1, qualitative testing of glue-injected meat with different concentrations was performed according to the steps of Example 9. Fig.10 In Example 9, the accuracy of near infrared and visible light quantitative detection of κ-carrageenan in pork reached 100%, precision 100%, recall rate 100%, F1-score 100%, and the classification model accuracy was as high as 100%, with high model accuracy. The results show that hyperspectral imaging technology combined with SVM model can be used to quickly and accurately detect pork products injected with different concentrations of κ-carrageenan without loss.

[0095] A pork product adulteration detection system based on hyperspectral imaging, comprising:

[0096] A dataset construction module is used to obtain hyperspectral image information of pork products adulterated with different concentrations of κ-carrageenan and construct a comparison set;

[0097] A first drawing module draws a spectrum diagram based on the hyperspectral image in the comparison set to obtain comparison data;

[0098] The second drawing module obtains the hyperspectral image of the pork product to be tested and draws the corresponding spectrum to obtain sample data;

[0099] The comparison and analysis module is used to analyze the sample data and the comparison data to obtain the detection result.

[0100] The contents of the above method embodiments are all applicable to the present system embodiments. The functions specifically implemented by the present system embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0101] A pork product adulteration detection device based on hyperspectral imaging:

[0102] at least one processor;

[0103] at least one memory for storing at least one program;

[0104] When the at least one program is executed by the at least one processor, the at least one processor implements the pork product adulteration detection method based on hyperspectral imaging as described above.

[0105] The contents of the above method embodiments are all applicable to the present device embodiments. The functions specifically implemented by the present device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0106] A storage medium stores processor-executable instructions, which are used to implement a method for detecting adulteration of pork products based on hyperspectral imaging when executed by a processor.

[0107] The contents of the above method embodiments are all applicable to the present storage medium embodiments. The functions specifically implemented by the present storage medium embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0108] The above is a specific description of the preferred implementation of the present invention, but the invention is not limited to the embodiments. Those skilled in the art may make various equivalent modifications or substitutions without violating the spirit of the present invention. These equivalent modifications or substitutions are all included in the scope defined by the claims of this application.

Claims

1. A method for detecting adulteration of pork products based on hyperspectral imaging, characterized in that: The following steps are involved: Obtain hyperspectral image information of pork products adulterated with different concentrations of κ-carrageenan and construct a comparison set; Drawing a spectrum diagram based on the hyperspectral image in the comparison set to obtain comparison data; Obtain a hyperspectral image of the pork product to be tested and draw a corresponding spectrum diagram to obtain sample data; The sample data and the comparison data are analyzed to obtain a test result.

2. The method for detecting adulteration of pork products based on hyperspectral imaging according to claim 1, characterized in that: The step of obtaining hyperspectral image information of pork products adulterated with different concentrations of κ-carrageenan and constructing a comparison set specifically includes: The adulteration concentration was set and pork meat products adulterated with different concentrations of κ-carrageenan were placed on a sample tray; Scanning the pork product based on a hyperspectral instrument and acquiring a corresponding image through data acquisition software to obtain a hyperspectral image; The adulteration concentration and the hyperspectral image are correspondingly associated to obtain a comparison set.

3. The method for detecting adulteration of pork products based on hyperspectral imaging according to claim 1, characterized in that: The step of drawing a spectrum graph based on the hyperspectral image in the comparison set to obtain comparison data specifically includes: Processing the hyperspectral image in the comparison set to remove background interference and obtain a region of interest; Based on the region of interest, generating reflectance values ​​of each wavelength of near infrared and visible light; The spectra were plotted using the reflectance values ​​of each wavelength of near-infrared and visible light to obtain comparative data.

4. The method for detecting adulteration of pork products based on hyperspectral imaging according to claim 1, characterized in that: The step of analyzing the sample data and the comparison data to obtain the test result specifically includes: According to the sample data and the comparison data, the characteristic peaks and troughs and their position shifts are observed, the differences are previewed, and the detection results are obtained.

5. The method for detecting adulteration of pork products based on hyperspectral imaging according to claim 4, characterized in that: The step of analyzing the sample data and the comparison data to obtain the test result further includes: The reflectance values ​​at each wavelength were used to perform partial least squares discriminant analysis to obtain the final test results.

6. The method for detecting adulteration of pork products based on hyperspectral imaging according to claim 4, characterized in that: The step of analyzing the sample data and the comparison data to obtain the test result further includes: The reflectance values ​​at each wavelength were used for SVM analysis to obtain the final detection results.

7. A pork product adulteration detection system based on hyperspectral imaging, characterized in that: include: A dataset construction module is used to obtain hyperspectral image information of pork products adulterated with different concentrations of κ-carrageenan and construct a comparison set; A first drawing module draws a spectrum diagram based on the hyperspectral image in the comparison set to obtain comparison data; The second drawing module obtains the hyperspectral image of the pork product to be tested and draws the corresponding spectrum to obtain sample data; The comparison and analysis module is used to analyze the sample data and the comparison data to obtain the detection result.

8. A pork product adulteration detection device based on hyperspectral imaging, characterized in that: include: at least one processor; at least one memory for storing at least one program; When the at least one program is executed by the at least one processor, the at least one processor implements the pork product adulteration detection method based on hyperspectral imaging as described in any one of claims 1 to 6.

Images