A method and system for quantitative analysis of gingerols in various kinds of ginger slices based on hyperspectral imaging technology

Abstract

This invention discloses a method and system for quantitative analysis of gingerol in various types of ginger slices based on hyperspectral imaging technology. The method uses five types of ginger slices—fresh ginger, small yellow ginger, galangal, turmeric, and large ginger—as the subjects. High-performance liquid chromatography (HPLC) is used to determine the content of gingerol-like substances as a reference value. Hyperspectral image data in both the visible-near-infrared and near-infrared bands are acquired using a hyperspectral imaging system. After white and dark calibration, the region of interest (ROI) spectral data is extracted. Savitzky-Golay convolution smoothing combined with derivative transformation is used for preprocessing. The preprocessed dual-band spectral data are then fused at a low level to form a fused spectral matrix. Using the fused spectral matrix as input and the HPLC determination value as output, a quantitative prediction model is established using partial least squares regression. This method achieves rapid and non-destructive quantitative analysis of gingerol content in various types of ginger slices and can be applied to ginger product quality control and production process monitoring.

Description

Technical Field

[0001] This invention belongs to the field of non-destructive testing technology for food, and in particular relates to a method and system for quantitative analysis of gingerol in various types of ginger slices based on hyperspectral imaging technology. Background Technology

[0002] Gingerol is the main material basis determining the spiciness and quality differences of ginger products. Its content and composition directly affect the flavor quality, processing suitability, and quality grade of different types of ginger products, such as ginger slices, ginger powder, ginger juice, and ginger extract. Therefore, establishing a high-efficiency quantitative analysis method for gingerol applicable to various types of ginger products is of great significance for achieving ginger product quality evaluation, processing control, and product grading. Currently, the detection of gingerol content mainly relies on traditional chemical analysis methods such as high-performance liquid chromatography, gas chromatography, and mass spectrometry. Although these methods have high detection accuracy, in practical applications, they usually require pretreatment steps such as sample crushing, extraction, purification, and separation, resulting in complex operation procedures, long detection cycles, high professional requirements for operators, and high detection costs. At the same time, traditional detection methods usually consume large amounts of organic solvents or other chemical reagents, and the samples after detection are difficult to recover and reuse, which is not conducive to the rapid screening of large batches of samples and the online detection needs in commercial production scenarios.

[0003] Existing detection methods still have shortcomings in terms of sample detection timeliness, making it difficult to provide real-time feedback on key quality indicators of ginger products during actual production or distribution, and failing to meet the needs of processing enterprises for rapid, non-destructive, and online quality monitoring. Although near-infrared spectroscopy, Raman spectroscopy, hyperspectroscopy, and other spectroscopic detection methods have been gradually applied to the quality evaluation of agricultural products and food, in existing technical solutions, single-spectral techniques often suffer from limited dimensions of chemical information obtained from samples, insufficient adaptability to different varieties and processing forms of samples, and weak stability and generalization ability of prediction models. As a result, the accuracy and applicability of gingerol quantitative analysis still need to be improved. The above-mentioned defects are caused by the fact that spectral information in different bands responds differently to different chemical components in the sample, and the spectral information contained in a single band is difficult to fully reflect the fingerprint characteristics of complex components in the sample. At the same time, the matrix differences of different types of ginger slices are large, and single-spectral models are difficult to maintain good predictive performance across different varieties, which brings difficulties to feature extraction and model generalization in the modeling process. To address the aforementioned issues, there is an urgent need for a detection method that can overcome the shortcomings of traditional chemical detection methods, such as complex operation, long time consumption, large reagent consumption, non-recoverable samples, insufficient acquisition of information from single-spectral detection, and limited model prediction performance, in order to achieve efficient, accurate, rapid, and non-destructive analysis of gingerol content and composition characteristics. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention proposes a method and system for quantitative analysis of gingerol in various types of ginger slices based on hyperspectral imaging technology, thereby resolving the issues existing in the prior art.

[0005] In a first aspect, to achieve the above objectives, the present invention provides a method for quantitative analysis of gingerol in various types of ginger slices based on hyperspectral imaging technology, comprising the following steps: Various types of ginger slices were prepared, and the content of gingerol-like substances in the samples was determined by high performance liquid chromatography as a reference value. A hyperspectral imaging system was used to acquire hyperspectral image data of the sample in the visible-near infrared and near infrared bands, and white calibration and dark calibration were performed on the image data. Spectral data of the region of interest is extracted from the calibrated hyperspectral image data. The extracted spectral data is normalized and then preprocessed using Savitzky-Golay convolution smoothing combined with derivative transformation. The preprocessed visible-near-infrared spectral data and near-infrared spectral data are fused at a low level and then stitched together to form a fused spectral matrix. Using the fused spectral matrix as input and the content of gingerol-like substances determined by the high performance liquid chromatography as output, a quantitative prediction model is established using partial least squares regression. Hyperspectral image data of ginger slices to be tested in the visible-near infrared and near infrared bands are collected, and after the same data preprocessing and low-level fusion processing, they are input into the quantitative prediction model to obtain the prediction results of gingerol content.

[0006] Optionally, the process of preparing multiple types of ginger slice samples includes: selecting ginger slices from five categories: fresh ginger, small yellow ginger, galangal, turmeric, and large ginger, and collecting sixty samples from each category; the process of determining the content of gingerol-like substances in the samples using high performance liquid chromatography includes: grinding fresh ginger slices after acquiring spectral images into a paste, weighing the ginger paste, adding methanol to make up to the volume, letting it stand, taking the supernatant, filtering it, and injecting it into a high performance liquid chromatograph, using a C18 column, with water-acetic acid and acetonitrile as the mobile phase, and determining the content of 6-gingerol at a detection wavelength of 270 nm.

[0007] Optionally, the process of acquiring hyperspectral image data of the sample in the visible-near-infrared and near-infrared bands using a hyperspectral imaging system includes: placing ginger slices on a quartz sample plate, vertically aligning the spectrometer probe with the sample surface, and acquiring data in reflection mode; the acquisition range for the visible-near-infrared band is 400 nm to 1000 nm, and the acquisition range for the near-infrared band is 900 nm to 1700 nm; white calibration and dark calibration are performed sequentially before acquisition, and the average frame count of the reference image is ten frames; the sample stage moving speed along the X-axis is set to 4.5 mm / s.

[0008] Optionally, the preprocessing of normalized spectral data using Savitzky-Golay convolutional smoothing combined with derivative transformation includes: selecting an effective band of 400 nm to 1000 nm for visible-near-infrared spectral data and an effective band of 900 nm to 1700 nm for near-infrared spectral data; filtering the spectral data within the effective bands using Savitzky-Golay convolutional smoothing, and then calculating the first or second derivative of the filtered spectral data; dividing the preprocessed samples into a calibration set and a prediction set, with a ratio of three to two.

[0009] Optionally, the process of performing low-level fusion of the preprocessed visible-near-infrared spectral data and near-infrared spectral data includes: normalizing the preprocessed visible-near-infrared spectral data and near-infrared spectral data to unify the dimensions of the two spectral data; directly splicing the normalized visible-near-infrared spectral data and near-infrared spectral data to form a fused spectral matrix; the fused spectral matrix contains 678 bands in the visible-near-infrared band and 510 bands in the near-infrared band, for a total of 1,188 wavelength points.

[0010] Optionally, the process of establishing a quantitative prediction model using the fused spectral matrix as input and partial least squares regression includes: using the fused spectral matrix as the model input matrix and the 6-gingerol content determined by high performance liquid chromatography as the model output variable; establishing a quantitative correction model for the correction set samples using the partial least squares regression algorithm; validating the established quantitative correction model using the prediction set samples to obtain the prediction set determination coefficient and prediction set root mean square error as evaluation indicators of the model's prediction performance; or establishing a quantitative prediction model using the random forest regression algorithm.

[0011] Secondly, the present invention also provides a system for quantitative analysis of gingerol in various types of ginger slices based on hyperspectral imaging technology, for implementing a method for quantitative analysis of gingerol in various types of ginger slices based on hyperspectral imaging technology, the system comprising: The reference value determination module is used to determine the content of gingerol-like substances in various types of ginger slices using high performance liquid chromatography, and to serve as a reference value. The hyperspectral image acquisition module is used to acquire hyperspectral image data of samples in the visible-near-infrared and near-infrared bands, and to perform white calibration and dark calibration on the image data; The spectral data extraction and preprocessing module is used to extract spectral data of the region of interest from the calibrated hyperspectral image data, normalize the extracted spectral data, and preprocess the normalized spectral data by using Savitzky-Golay convolution smoothing combined with derivative transformation. The data fusion module is used to perform low-level fusion of preprocessed visible-near-infrared spectral data and near-infrared spectral data, and splice them together to form a fused spectral matrix. The model building module is used to establish a quantitative prediction model by taking the fused spectral matrix as input, the content of gingerol-like substances determined by the high performance liquid chromatography as output, and partial least squares regression. The prediction module is used to collect hyperspectral image data of ginger slices in the visible-near infrared and near infrared bands. After the same data preprocessing and low-level fusion processing, the data is input into the quantitative prediction model to obtain the prediction results of gingerol content.

[0012] Thirdly, the present invention also provides a computer terminal device, comprising: One or more processors; A memory, coupled to the processor, for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the steps of the method for quantitative analysis of gingerol in various types of ginger slices based on hyperspectral imaging technology in the first aspect described above.

[0013] Fourthly, the present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, it implements the steps of the method for quantitative analysis of gingerol in various types of ginger slices based on hyperspectral imaging technology in the first aspect described above.

[0014] Fifthly, the present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the method for quantitative analysis of gingerol in various types of ginger slices based on hyperspectral imaging technology described in the first aspect above.

[0015] Compared with the prior art, the present invention has the following advantages and technical effects: This invention provides a method and system for quantitative analysis of gingerol in various types of ginger slices based on hyperspectral imaging technology. It overcomes the shortcomings of traditional chemical detection methods, such as complex operation, long processing time, high reagent consumption, and non-recoverable samples. It eliminates the need for pretreatment steps such as sample crushing, extraction, purification, and separation, enabling rapid prediction of gingerol content in ginger slices. This reduces organic solvent consumption, preserves sample integrity and recyclability, and significantly shortens the detection cycle. Furthermore, this invention overcomes the limitation of single-spectral techniques in obtaining limited chemical information by fusing visible-near-infrared hyperspectral imaging with near-infrared hyperspectral imaging. It fully utilizes the complementarity of information from different spectral bands, extracting more comprehensive feature information and enhancing the model's adaptability and generalization ability to different types of ginger slices. The quantitative prediction model constructed by this invention can be applied to ginger product quality evaluation, processing control, and product grading, meeting the needs of efficient, accurate, rapid, and non-destructive analysis of gingerol content and composition characteristics in actual production. Attached Figure Description

[0016] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings: Figure 1 The figures show the average reflectance spectra of five types of ginger slices in the visible-near-infrared and near-infrared bands, respectively, according to embodiments of the present invention. Figure 2 This is a comparison of the spectra before and after preprocessing with Savitzky-Golay convolution smoothing combined with derivative transformation in an embodiment of the present invention; Figure 3 This is a comparison chart of the best modeling results of visible-near-infrared spectra and near-infrared spectra after different preprocessing methods in an embodiment of the present invention; Figure 4 This is a comparison chart of the best modeling results under low-level data fusion and mid-level data fusion strategies in this embodiment of the invention; Figure 5 This is a flowchart illustrating a method for quantitative analysis of gingerol in various types of ginger slices based on hyperspectral imaging technology, according to an embodiment of the present invention. Detailed Implementation

[0017] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0018] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0019] Example 1 like Figure 5 As shown, this embodiment provides a method for quantitative analysis of gingerol in various types of ginger slices based on hyperspectral imaging technology, including: Various types of ginger slices were prepared, and the content of gingerol-like substances in the samples was determined by high performance liquid chromatography as a reference value. A hyperspectral imaging system was used to acquire hyperspectral image data of the sample in the visible-near infrared and near infrared bands, and white calibration and dark calibration were performed on the image data. Spectral data of the region of interest is extracted from the calibrated hyperspectral image data. The extracted spectral data is normalized and then preprocessed using Savitzky-Golay convolution smoothing combined with derivative transformation. The preprocessed visible-near-infrared spectral data and near-infrared spectral data are fused at a low level and then stitched together to form a fused spectral matrix. Using the fused spectral matrix as input and the content of gingerol-like substances determined by the high performance liquid chromatography as output, a quantitative prediction model is established using partial least squares regression. Hyperspectral image data of ginger slices to be tested in the visible-near infrared and near infrared bands are collected, and after the same data preprocessing and low-level fusion processing, they are input into the quantitative prediction model to obtain the prediction results of gingerol content.

[0020] Furthermore, the process of preparing various types of ginger slice samples includes: selecting ginger slices from five categories: fresh ginger, small yellow ginger, galangal, turmeric, and large ginger, and collecting sixty samples from each category; the process of determining the content of gingerol-like substances in the samples using high performance liquid chromatography includes: grinding fresh ginger slices after acquiring spectral images into a paste, weighing the ginger paste, adding methanol to make up to the volume, letting it stand, taking the supernatant, filtering it, and injecting it into a high performance liquid chromatograph, using a C18 column, with water-acetic acid and acetonitrile as the mobile phase, and determining the content of 6-gingerol at a detection wavelength of 270 nm.

[0021] Specifically, the implementation process of this embodiment includes: HPLC determination of gingerol-like substances in Chinese ginger: Sixty samples were collected from each category, yielding a total of 300 visible and near-infrared (VIN) images and 300 NIN spectral data. Fresh ginger slices, after spectral image acquisition, were ground into a paste using a food-grade blender. 1.00 g of ginger paste was accurately weighed and placed in a 50 mL volumetric flask. Methanol was added as the extraction solvent, and the mixture was brought to volume. The mixture was allowed to stand at room temperature for 12 hours. The extract was transferred to a centrifuge tube, and the supernatant was filtered through a 0.22 μm syringe organic filter membrane. The filtrate was collected and injected into a liquid chromatography vial for HPLC analysis. The contents of 6-gingerol, 6-shogaol, 8-gingerol, and 8-gingerol were determined. The experiment used a high-performance liquid chromatograph equipped with a UV detector and a C18 column (4.6 × 250 mm, 5 μm). Mobile phase A was water-acetic acid (99:1), and mobile phase B was acetonitrile, with a mobile phase ratio of 65:35. The detection wavelength was 270 nm, and the detection time was 35 minutes.

[0022] Furthermore, the process of acquiring hyperspectral image data of the sample in the visible-near-infrared and near-infrared bands using a hyperspectral imaging system includes: placing ginger slices on a quartz sample plate, vertically aligning the spectrometer probe with the sample surface, and acquiring data using reflection mode; the acquisition range for the visible-near-infrared band is 400nm to 1000nm, and the acquisition range for the near-infrared band is 900nm to 1700nm; white calibration and dark calibration are performed sequentially before acquisition, and the average frame count of the reference image is ten frames; the sample stage moving speed along the X-axis is set to 4.5mm / s.

[0023] Specifically, the implementation process of this embodiment includes: Hyperspectral image data acquisition was performed using a hyperspectral imaging system. Experimental parameters were set as follows: image spatial resolution of 640×512 pixels, camera exposure time of 43ms, gain of 0, frame rate of 20fps (actual frame rate of 20.2fps), CCD sensor operating temperature of 29.63℃, operating mode of Mono, and scanning method of Vertical push-broom. Pre-treated ginger slices were placed on a quartz sample plate, and the spectrometer probe was adjusted to be perpendicularly aligned with the sample surface. Spectral data acquisition was then performed using reflectance mode. Before data acquisition, white calibration (Ref-White) and dark calibration (Ref-Dark) were performed sequentially, with an average of 10 frames per second for the reference image. The sample stage was controlled via an RS232C interface (COM3), with a movement speed of 4.5mm / s along the X-axis to ensure matching with the camera frame rate and achieve stable push-broom imaging. This embodiment uses two bands, Vis-NIR and NIR, for hyperspectral image acquisition: Vis-NIR band: 400-1000nm, spectral resolution 2.8nm, a total of 678 bands; NIR band: 900-1700nm, spectral resolution 2.6nm, a total of 510 bands.

[0024] Furthermore, the preprocessing of the normalized spectral data using Savitzky-Golay convolutional smoothing combined with derivative transformation includes: selecting an effective band of 400nm to 1000nm for visible-near-infrared spectral data and an effective band of 900nm to 1700nm for near-infrared spectral data; filtering the spectral data within the effective bands using Savitzky-Golay convolutional smoothing, and then calculating the first or second derivative of the filtered spectral data; dividing the preprocessed samples into a calibration set and a prediction set, with a ratio of three to two.

[0025] Specifically, the implementation process of this embodiment includes: To ensure the data reliability and modeling quality of the constructed model, this embodiment performs systematic preprocessing operations and feature variable screening on the original hyperspectral data. For example, Figure 1 The visible and near-infrared data selected an effective wavelength range of 400-1000 nm, while the near-infrared spectral data selected an effective wavelength range of 900-1700 nm to remove edge noise and invalid information, retaining only spectral signals with practical value for modeling. To address the issues of incompatible dimensions and inconsistent signal amplitudes between different spectral data, normalization processing was performed on the spectra of different modes. For example... Figure 2 As shown, Savitzky-Golay convolution smoothing combined with derivative transformation is used for preprocessing to address random noise, baseline drift, and light scattering interference present in the original hyperspectral image. Figure 3 This is a schematic diagram illustrating the optimal results of single-spectrum preprocessing (for visible and near-infrared (VIS-NIR) spectra, after processing with different preprocessing methods, the optimal preprocessing method is SG-2ndDer, and the modeling effect is Rc). 2 =0.9717, Rp 2 =0.9527, RMSEC =0.3432, RMSEP =0.4504, RPD = 4.5748; For near-infrared spectroscopy (NIR), the optimal preprocessing method Savitzky-Golay was selected, and the modeling effect was Rc 2 = 0.9583, Rp 2=0.9227, RMSEC=0.5833, RMSEP=0.7989, RPD=2.5791). This method, through polynomial fitting within a moving window, can significantly suppress noise levels while preserving core spectral features. Combining the inherent characteristics of the spectrum with the model optimization objective, the first and second derivatives are selected for processing: the first derivative can effectively correct baseline drift, enhance absorption peak contours, and improve model stability; the second derivative can distinguish overlapping absorption peaks and subtle spectral differences, achieving more refined feature information mining. To ensure sufficient representativeness and uniform data distribution of the modeling samples, the samples are divided into a calibration set (approximately 60%) and a prediction set (approximately 40%), with a ratio of 3:2. The model outputs are sorted by size. Every 5 consecutive samples are used as a unit, with the 1st, 3rd, and 5th samples included in the calibration set, and the 2nd and 4th samples included in the prediction set, so that both the calibration and prediction sets can reflect the spectral differences of the samples in a balanced manner.

[0026] Furthermore, the low-level fusion process of the preprocessed visible-near-infrared spectral data and near-infrared spectral data includes: normalizing the preprocessed visible-near-infrared spectral data and near-infrared spectral data to unify the dimensions of the two spectral data; directly splicing the normalized visible-near-infrared spectral data and near-infrared spectral data to form a fused spectral matrix; the fused spectral matrix contains 678 bands in the visible-near-infrared band and 510 bands in the near-infrared band, for a total of 1,188 wavelength points.

[0027] Specifically, the implementation process of this embodiment includes: Figure 4 The diagram illustrates the optimal results of the primary and intermediate fusion models (Primary fusion model: Under the low-level data fusion strategy, Vis-NIR and NIR spectra are directly spliced ​​together after optimal preprocessing to form a fusion spectral matrix (a total of 1188 wavelength points). After normalization, the MFO variable selection method-PLS model exhibits the best predictive performance: calibration set coefficient of determination Rc2=0.9802, prediction set coefficient of determination Rp2=0.9733, calibration set root mean square error RMSEC=0.2875, and prediction set root mean square error RMSEP=0.3351; The core of the intermediate data fusion strategy lies in using the MFO method to extract key variables from Vis-NIR and NIR spectra respectively. Vis-NIR selects 366 wavelength points, and NIR selects 257 wavelength points, for a total of 623 variables, which are then normalized to construct the intermediate fusion matrix). Low-level data fusion, also known as raw data layer fusion, involves directly splicing raw or preprocessed spectral data obtained from different detection methods according to certain rules. The spectra are normalized before splicing to unify the dimensions of the spectra before data integration. This avoids spectral intensity masking the characteristics between spectra, constructing a new dataset that serves as model input. In this embodiment, visible and near-infrared data are directly spliced ​​together to construct a fused spectral matrix, preserving the original information of multi-source spectra to the greatest extent possible. This method can fully preserve spectral details, expand feature dimensions, and effectively enhance the robustness of the model.

[0028] Intermediate data fusion, also known as feature set fusion, involves first filtering variables from multiple data sources separately, and then fusing the extracted feature sets. This process removes redundant and irrelevant variables, strengthens key features associated with the target component, and thus improves the model's prediction accuracy. This embodiment employs the Variable Combination Moth to a Flame (MFO) optimization search algorithm, which is primarily a global optimization variable filtering method. It retains most spectral variables, has a large number of feature variables, and relatively completely preserves the corresponding chemical information. The MFO method can filter feature variables from visible-near-infrared (VIS-NIR) and near-infrared (NIR) spectral datasets separately, and then concatenate the filtered feature variables to form a fused feature set, which serves as the input parameters for the intermediate fusion model.

[0029] Furthermore, the process of establishing a quantitative prediction model using the fused spectral matrix as input and partial least squares regression includes: using the fused spectral matrix as the model input matrix and the 6-gingerol content determined by high performance liquid chromatography as the model output variable; establishing a quantitative correction model for the correction set samples using the partial least squares regression algorithm; validating the established quantitative correction model using the prediction set samples to obtain the prediction set determination coefficient and prediction set root mean square error as evaluation indicators of model prediction performance; or establishing a quantitative prediction model using the random forest regression algorithm.

[0030] Specifically, the implementation process of this embodiment includes: The visible-near-infrared and near-infrared dual spectra acquired by hyperspectral imaging technology are used as the input source of the matrix, and the content of gingerol-like substances 6-gingerol in ginger slices measured by high performance liquid chromatography (HPLC) is used as the quantitative output source for quantitative analysis of visible-near-infrared, near-infrared single spectral models and multispectral fusion models.

[0031] Example 2 Based on the same general inventive concept, this invention also provides a system for quantitative analysis of gingerol in various types of ginger slices using hyperspectral imaging technology. The system for quantitative analysis of gingerol in various types of ginger slices using hyperspectral imaging technology provided by this invention is described below. The system described below can be referred to in conjunction with the method described above for quantitative analysis of gingerol in various types of ginger slices using hyperspectral imaging technology. The system includes: The reference value determination module is used to determine the content of gingerol-like substances in various types of ginger slices using high performance liquid chromatography, and to serve as a reference value. The hyperspectral image acquisition module is used to acquire hyperspectral image data of samples in the visible-near-infrared and near-infrared bands, and to perform white calibration and dark calibration on the image data; The spectral data extraction and preprocessing module is used to extract spectral data of the region of interest from the calibrated hyperspectral image data, normalize the extracted spectral data, and preprocess the normalized spectral data by using Savitzky-Golay convolution smoothing combined with derivative transformation. The data fusion module is used to perform low-level fusion of preprocessed visible-near-infrared spectral data and near-infrared spectral data, and splice them together to form a fused spectral matrix. The model building module is used to establish a quantitative prediction model by taking the fused spectral matrix as input, the content of gingerol-like substances determined by the high performance liquid chromatography as output, and partial least squares regression. The prediction module is used to collect hyperspectral image data of ginger slices in the visible-near infrared and near infrared bands. After the same data preprocessing and low-level fusion processing, the data is input into the quantitative prediction model to obtain the prediction results of gingerol content.

[0032] Example 3 (equivalent to Example 2 above) In this embodiment, a computer terminal device is provided, including: One or more processors; A memory, coupled to the processor, for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the steps of the above-described method for quantitative analysis of gingerol in various types of ginger slices based on hyperspectral imaging technology.

[0033] In this embodiment, a computer-readable storage medium is also provided, on which a computer program is stored. When the computer program is executed by a processor, it implements the steps of the above-described method for quantitative analysis of gingerol in various types of ginger slices based on hyperspectral imaging technology.

[0034] In this embodiment, a computer program product is also provided, including a computer program that, when executed by a processor, implements the steps of the above-described method for quantitative analysis of gingerol in various types of ginger slices based on hyperspectral imaging technology.

[0035] This invention provides a method and system for quantitative analysis of gingerol in various types of ginger slices based on hyperspectral imaging technology. It overcomes the shortcomings of traditional chemical detection methods, such as complex operation, long processing time, high reagent consumption, and non-recoverable samples. It eliminates the need for pretreatment steps such as sample crushing, extraction, purification, and separation, enabling rapid prediction of gingerol content in ginger slices. This reduces organic solvent consumption, preserves sample integrity and recyclability, and significantly shortens the detection cycle. Furthermore, this invention overcomes the limitation of single-spectral techniques in obtaining limited chemical information by fusing visible-near-infrared hyperspectral imaging with near-infrared hyperspectral imaging. It fully utilizes the complementarity of information from different spectral bands, extracting more comprehensive feature information and enhancing the model's adaptability and generalization ability to different types of ginger slices. The quantitative prediction model constructed by this invention can be applied to ginger product quality evaluation, processing control, and product grading, meeting the needs of efficient, accurate, rapid, and non-destructive analysis of gingerol content and composition characteristics in actual production.

[0036] The above are merely preferred embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for quantitative analysis of gingerol in various types of ginger slices based on hyperspectral imaging technology, characterized in that, Includes the following steps: Various types of ginger slices were prepared, and the content of gingerol-like substances in the samples was determined by high performance liquid chromatography as a reference value. A hyperspectral imaging system was used to acquire hyperspectral image data of the sample in the visible-near infrared and near infrared bands, and white calibration and dark calibration were performed on the image data. Spectral data of the region of interest is extracted from the calibrated hyperspectral image data. The extracted spectral data is normalized and then preprocessed using Savitzky-Golay convolution smoothing combined with derivative transformation. The preprocessed visible-near-infrared spectral data and near-infrared spectral data are fused at a low level and then stitched together to form a fused spectral matrix. Using the fused spectral matrix as input and the content of gingerol-like substances determined by the high performance liquid chromatography as output, a quantitative prediction model is established using partial least squares regression. Hyperspectral image data of ginger slices to be tested in the visible-near infrared and near infrared bands are collected, and after the same data preprocessing and low-level fusion processing, they are input into the quantitative prediction model to obtain the prediction results of gingerol content.

2. The method according to claim 1, characterized in that, The process of preparing various types of ginger slice samples includes: selecting ginger slices from five categories: fresh ginger, small yellow ginger, galangal, turmeric, and large ginger, and collecting sixty samples from each category; the process of determining the content of gingerol-like substances in the samples using high performance liquid chromatography includes: grinding fresh ginger slices into a paste after the spectral images have been collected, weighing the ginger paste, adding methanol to make up to the volume, letting it stand, taking the supernatant, filtering it, and injecting it into a high performance liquid chromatograph, using a C18 column, with water-acetic acid and acetonitrile as the mobile phase, and determining the content of 6-gingerol at a detection wavelength of 270 nm.

3. The method according to claim 1, characterized in that, The process of acquiring hyperspectral image data of the sample in the visible-near-infrared and near-infrared bands using a hyperspectral imaging system includes: placing ginger slices on a quartz sample plate, vertically aligning the spectrometer probe with the sample surface, and acquiring data in reflection mode; the acquisition range for the visible-near-infrared band is 400nm to 1000nm, and the acquisition range for the near-infrared band is 900nm to 1700nm; white calibration and dark calibration are performed sequentially before acquisition, and the average frame rate of the reference image is ten frames; the sample stage moving speed along the X-axis is set to 4.5mm / s.

4. The method according to claim 1, characterized in that, The preprocessing of normalized spectral data using Savitzky-Golay convolutional smoothing combined with derivative transformation includes: selecting an effective band of 400 nm to 1000 nm for visible-near-infrared spectral data and an effective band of 900 nm to 1700 nm for near-infrared spectral data; filtering the spectral data within the effective bands using Savitzky-Golay convolutional smoothing, and then calculating the first or second derivative of the filtered spectral data; dividing the preprocessed samples into a calibration set and a prediction set, with a ratio of three to two.

5. The method according to claim 1, characterized in that, The process of low-level fusion of preprocessed visible-near-infrared spectral data and near-infrared spectral data includes: normalizing the preprocessed visible-near-infrared spectral data and near-infrared spectral data to unify the dimensions of the two spectral data; directly splicing the normalized visible-near-infrared spectral data and near-infrared spectral data to form a fused spectral matrix; the fused spectral matrix contains 678 bands in the visible-near-infrared band and 510 bands in the near-infrared band, for a total of 1,188 wavelength points.

6. The method according to claim 1, characterized in that, The process of establishing a quantitative prediction model using a fused spectral matrix as input and partial least squares regression includes: using the fused spectral matrix as the model input matrix and the 6-gingerol content determined by high performance liquid chromatography as the model output variable; establishing a quantitative correction model for the correction set samples using the partial least squares regression algorithm; validating the established quantitative correction model using the prediction set samples to obtain the prediction set determination coefficient and prediction set root mean square error as evaluation indicators of the model's prediction performance; or establishing a quantitative prediction model using the random forest regression algorithm.

7. A system for quantitative analysis of gingerol in various types of ginger slices based on hyperspectral imaging technology, characterized in that, The system for implementing the method of any one of claims 1-6 comprises: The reference value determination module is used to determine the content of gingerol-like substances in various types of ginger slices using high performance liquid chromatography, and to serve as a reference value. The hyperspectral image acquisition module is used to acquire hyperspectral image data of samples in the visible-near-infrared and near-infrared bands, and to perform white calibration and dark calibration on the image data; The spectral data extraction and preprocessing module is used to extract spectral data of the region of interest from the calibrated hyperspectral image data, normalize the extracted spectral data, and preprocess the normalized spectral data by using Savitzky-Golay convolution smoothing combined with derivative transformation. The data fusion module is used to perform low-level fusion of preprocessed visible-near-infrared spectral data and near-infrared spectral data, and splice them together to form a fused spectral matrix. The model building module is used to establish a quantitative prediction model by taking the fused spectral matrix as input, the content of gingerol-like substances determined by the high performance liquid chromatography as output, and partial least squares regression. The prediction module is used to collect hyperspectral image data of ginger slices in the visible-near infrared and near infrared bands. After the same data preprocessing and low-level fusion processing, the data is input into the quantitative prediction model to obtain the prediction results of gingerol content.

8. A computer terminal device, characterized in that, include: One or more processors; A memory, coupled to the processor, for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors perform the steps of the method as described in any one of claims 1-6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1-6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1-6.

Images