Method, system and apparatus for determining acrylamide content profile in roast coffee
By using hyperspectral imaging technology and a multiple linear regression model, the high cost and time-consuming nature of traditional methods for detecting acrylamide content in roasted coffee have been solved, achieving efficient and low-cost acrylamide content detection and supporting food safety testing.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- ZHEJIANG ACADEMY OF AGRICULTURE SCIENCES
- Filing Date
- 2023-05-18
- Publication Date
- 2026-04-24
AI Technical Summary
Traditional liquid chromatography-tandem mass spectrometry (LC-MS/MS) and gas chromatography-tandem mass spectrometry (GC-MS/MS) methods for detecting acrylamide content in roasted coffee are costly, labor-intensive, and time-consuming, making them unsuitable for the needs of mass food production.
Hyperspectral imaging technology was used to obtain standard white and blackboard images for image correction, and the spectral reflectance curve of each pixel in the hyperspectral corrected image of roasted coffee was determined. The acrylamide content was predicted using a multiple linear regression model, and an acrylamide content distribution map was constructed.
It improves the efficiency of acrylamide content detection, reduces the workload and cost of detection, and provides technical support for large-scale food safety testing.
Smart Images
Figure CN116359143B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of acrylamide content technology, and in particular to a method, system and equipment for determining the acrylamide content distribution in roasted coffee. Background Technology
[0002] During the coffee roasting process, reducing sugars and free asparagine undergo a Maillard reaction at high temperatures to form acrylamide. Acrylamide is a white organic compound that is carcinogenic to humans; therefore, the detection of acrylamide content is of great significance to human health.
[0003] Traditional methods for detecting acrylamide content in food include liquid chromatography-tandem mass spectrometry (LC-MS / MS) and gas chromatography-tandem mass spectrometry (GC-MS / MS). However, these two methods are costly, labor-intensive, and time-consuming. These inefficient methods for detecting acrylamide content are difficult to meet the needs of mass food production. Summary of the Invention
[0004] The purpose of this invention is to provide a method, system, and equipment for determining the acrylamide content distribution in roasted coffee, which can improve the efficiency of acrylamide content detection and reduce the workload and cost of acrylamide content detection, providing technical support for large-scale food safety testing.
[0005] To achieve the above objectives, the present invention provides the following solution:
[0006] A method for determining the acrylamide content distribution in roasted coffee, comprising:
[0007] Acquire standard whiteboard images, standard blackboard images, and the raw hyperspectral images of the coffee to be tested;
[0008] Using the standard whiteboard image and the standard blackboard image, the original hyperspectral image of the coffee to be tested is corrected to obtain the corrected hyperspectral image of the coffee to be tested.
[0009] Determine the full-spectral reflectance curve of each pixel in the hyperspectral corrected image of the coffee to be tested;
[0010] Determine any pixel as the current pixel;
[0011] The spectral reflectance of multiple feature bands is obtained from the spectral reflectance curve of the current pixel across the entire spectral band.
[0012] The spectral reflectance of multiple feature bands of the current pixel is input into the acrylamide content prediction model to obtain the acrylamide content of the roasted coffee to be tested at the current pixel. The coefficients of the acrylamide content prediction model are obtained by fitting a multiple linear regression model with the acrylamide content of the roasted coffee sample as the dependent variable and the average spectral reflectance of multiple feature bands of the roasted coffee sample as the independent variable.
[0013] Update the current pixel and return to the step "Obtain the spectral reflectance of multiple feature bands from the spectral reflectance curve of the full spectrum of the current pixel" until all pixels are traversed to obtain the acrylamide content of the roasted coffee sample at each pixel in the hyperspectral corrected image.
[0014] A distribution map of acrylamide content is constructed based on the acrylamide content of the roasted coffee sample at each pixel.
[0015] Optionally, the standard whiteboard image has a spectral reflectance of 100%; the standard blackboard image has a spectral reflectance of 0.
[0016] Optionally, the hyperspectral corrected image is:
[0017]
[0018] Among them, I calibrated For hyperspectral corrected images; I raw This is the original hyperspectral image; I white A standard whiteboard image; I dark Standard blackboard image.
[0019] Optionally, after determining the spectral reflectance curve of each pixel in the hyperspectral corrected image of the coffee to be tested, the method further includes:
[0020] The spectral reflectance curve of each pixel across the entire spectral band is smoothed.
[0021] Optionally, the process may include, before acquiring the standard whiteboard image, the standard blackboard image, and the original hyperspectral image of the coffee to be tested, the following:
[0022] The acrylamide content in multiple roasted coffee samples was determined using liquid chromatography-tandem mass spectrometry.
[0023] Multiple roasted coffee samples were acquired as raw hyperspectral images of the samples.
[0024] Using the standard whiteboard image and the standard blackboard image, each sample's original hyperspectral image is corrected to obtain multiple corrected hyperspectral images of the samples;
[0025] Determine the average spectral reflectance curve of each sample hyperspectral corrected image; the average spectral reflectance curve is the average value of the spectral reflectance curves of all pixels in the corresponding sample hyperspectral corrected image across the entire spectral band.
[0026] Based on the average spectral reflectance curve, the average spectral reflectance of multiple characteristic bands corresponding to each roasted coffee sample is determined.
[0027] Using the acrylamide content of roasted coffee samples as the dependent variable and the average spectral reflectance of multiple characteristic bands corresponding to the roasted coffee samples as the independent variable, a multiple linear regression model was fitted to obtain an acrylamide content prediction model.
[0028] Optionally, after determining the average spectral reflectance profile of each sample's hyperspectral corrected image, the method further includes:
[0029] The average spectral reflectance curve is smoothed.
[0030] A system for determining the acrylamide content distribution in roasted coffee, comprising:
[0031] The image acquisition module is used to acquire standard whiteboard images, standard blackboard images, and the original hyperspectral image of the roasted coffee to be tested.
[0032] The hyperspectral correction image acquisition module is used to correct the original hyperspectral image of the roasted coffee to be tested using the standard whiteboard image and the standard blackboard image, so as to obtain the hyperspectral correction image of the roasted coffee to be tested.
[0033] The spectral reflectance curve determination module is used to determine the full-spectral reflectance curve of each pixel in the hyperspectral correction image of the roasted coffee to be tested.
[0034] The current pixel determination module is used to determine any pixel as the current pixel.
[0035] The feature band spectral reflectance determination module is used to obtain the spectral reflectance of multiple feature bands from the spectral reflectance curve of the current pixel across the entire spectral band.
[0036] The acrylamide content determination module is used to input the spectral reflectance of multiple characteristic bands of the current pixel into the acrylamide content prediction model to obtain the acrylamide content of the roasted coffee to be tested at the current pixel; the coefficients of the acrylamide content prediction model are obtained by fitting a multiple linear regression model with the acrylamide content of the roasted coffee sample as the dependent variable and the average spectral reflectance of multiple characteristic bands of the roasted coffee sample as the independent variable.
[0037] The multi-pixel acrylamide content determination module is used to update the current pixel and call the characteristic band spectral reflectance determination module until all pixels are traversed to obtain the acrylamide content of the roasted coffee sample at each pixel in the hyperspectral corrected image.
[0038] The acrylamide content distribution determination module is used to construct an acrylamide content distribution map based on the acrylamide content of the roasted coffee sample at each pixel.
[0039] An electronic device includes a memory and a processor, the memory storing a computer program, and the processor running the computer program to cause the electronic device to perform the method for determining the acrylamide content distribution in roasted coffee.
[0040] Optionally, the memory is a readable storage medium.
[0041] According to specific embodiments provided by the present invention, the present invention discloses the following technical effects:
[0042] This invention provides a method, system, and device for determining the acrylamide content distribution in roasted coffee. Using standard whiteboard and blackboard images, the original hyperspectral image of the roasted coffee to be tested is corrected to obtain a corrected hyperspectral image. The spectral reflectance curve of the entire spectral band of each pixel in the corrected hyperspectral image is determined. The spectral reflectance of multiple characteristic bands is obtained from the spectral reflectance curve of the current pixel. The spectral reflectance of the multiple characteristic bands of the current pixel is input into an acrylamide content prediction model to obtain the acrylamide content of the roasted coffee at the current pixel. An acrylamide content distribution map is constructed based on the acrylamide content of the roasted coffee sample at each pixel. This invention utilizes the principle of multiple linear regression to construct an acrylamide content prediction model, which can improve the efficiency of acrylamide content detection and reduce the workload and cost of acrylamide content detection, providing technical support for large-scale food safety testing. Attached Figure Description
[0043] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0044] Figure 1 This is a flowchart of the method for determining the acrylamide content distribution in roasted coffee in Embodiment 1 of the present invention;
[0045] Figure 2This is a schematic diagram illustrating the principle of the method for determining the acrylamide content distribution in roasted coffee in Embodiment 1 of the present invention.
[0046] Figure 3 This is a schematic diagram illustrating the acquisition of the original hyperspectral image of roasted coffee in Embodiment 1 of the present invention;
[0047] Figure 4 This is a spectral reflectance curve of the roasted coffee sample in Example 1 of the present invention;
[0048] Figure 5 This is a schematic diagram of the characteristic bands selected in Embodiment 1 of the present invention;
[0049] Figure 6 This is a schematic diagram of the prediction results of acrylamide content in roasted coffee based on the characteristic band MLR model in Embodiment 1 of the present invention;
[0050] Figure 7 This is a visual distribution diagram of acrylamide content in dark roast coffee in Example 1 of the present invention;
[0051] Figure 8 This is a visual distribution diagram of acrylamide content in medium-roasted coffee in Example 1 of the present invention. Detailed Implementation
[0052] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0053] The purpose of this invention is to provide a method, system, and equipment for determining the acrylamide content distribution in roasted coffee, which can improve the efficiency of acrylamide content detection and reduce the workload and cost of acrylamide content detection, providing technical support for large-scale food safety testing.
[0054] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.
[0055] Example 1
[0056] like Figure 2This embodiment provides a method for determining the acrylamide content distribution in roasted coffee. Based on the spectral reflectance information of an image, a predictive model for the acrylamide content of the target substance is established by combining traditional physicochemical detection methods. A prediction equation is fitted to obtain the model, and then the model is inverted based on the spatial feature information of the image to achieve rapid and visualized detection of acrylamide content in roasted coffee. The method involves: acquiring near-infrared hyperspectral images of roasted coffee samples using a hyperspectral imager; processing the acquired hyperspectral images, including image correction using a black and white board, extracting spectral reflectance values from the corrected hyperspectral images, smoothing and optimizing the spectral reflectance values, and dividing them into a modeling set and a prediction set at a 2:1 ratio; extracting feature bands sensitive to the target substance based on a continuous projection algorithm; using the selected feature bands as input, establishing a multiple linear regression (MLR) model based on the modeling set samples; based on the established regression model, inputting the spectral reflectance values of the prediction set samples to obtain the Y value (acrylamide value) of the prediction set samples; and using the spatial information of the hyperspectral corrected image, combined with the fitted equation, inverting the acrylamide value corresponding to each pixel in the image, with different values corresponding to different colors, ultimately obtaining a visualized image showing the distribution of acrylamide.
[0057] Specifically, such as Figure 1 As shown, this embodiment provides a method for determining the acrylamide content distribution in roasted coffee, including:
[0058] Step 101: Acquire the standard whiteboard image, the standard blackboard image, and the original hyperspectral image of the coffee to be tested. The standard whiteboard image has a spectral reflectance of 100%; the standard blackboard image has a spectral reflectance of 0.
[0059] like Figure 3 Hyperspectral images of roasted coffee were acquired using a hyperspectral imaging system with a spectral range of 874.41–1733.91 nm. The system consisted of an N17E Specim imaging spectrometer, a digital CCD camera, a conveyor belt, and a pair of 150W halogen light sources. The system scanned the roasted coffee samples on the moving conveyor belt line by line. Simultaneously, images of a standard white plate with 100% spectral reflectance were acquired. white ; Cover the lens cap and turn off the light source to obtain a standard blackboard image I with a spectral reflectance of 0. dark Then, the acquired raw hyperspectral image I raw After correction, the hyperspectral corrected image I is obtained. calibrated .
[0060] Step 102: Using standard whiteboard and standard blackboard images, correct the original hyperspectral image of the roasted coffee to be tested, obtaining the corrected hyperspectral image of the roasted coffee. The corrected hyperspectral image is as follows:
[0061]
[0062] Among them, I calibrated For hyperspectral corrected images; I raw This is the original hyperspectral image; I white A standard whiteboard image; I dark Standard blackboard image.
[0063] Step 103: Determine the spectral reflectance curve of each pixel in the hyperspectral corrected image of the coffee to be tested, covering the entire spectral band.
[0064] Within the wavelength range of 874.41–1733.91 nm, the original spectral reflectance was smoothed using the Savitzky-Golay method. The spectral reflectance after Savitzky-Golay smoothing is shown below. Figure 4 As shown. All subsequent analyses were based on Savitzky-Golay smoothed data. Peaks and troughs were observed throughout the spectral region, attributed to molecular vibrations of chemical groups.
[0065] Step 104: Determine any pixel as the current pixel.
[0066] Step 105: Obtain the spectral reflectance of multiple feature bands from the spectral reflectance curve of the current pixel across the entire spectral band.
[0067] Five characteristic bands (888, 1123, 1456, 1636, and 1734 nm) were selected using the Continuous Projection Algorithm (SPA). Wavenumbers with larger positive or loading values were considered significant for the Y variable and were therefore identified as valid bands, indicating that these bands might contain more useful information for acrylamide detection. These valid wavenumbers not only reduced computational complexity but also simplified the model. An MLR model was then built based on the selected specific bands. The test values (t-values) for the selected bands are shown below. Figure 5 As shown. This can be used to evaluate the performance of each variable; the larger the absolute t-value, the more useful the band. In this study, we can see that 888nm performed best among the five selected bands.
[0068] Step 106: Input the spectral reflectance of multiple feature bands of the current pixel into the acrylamide content prediction model to obtain the acrylamide content of the roasted coffee to be tested at the current pixel; the coefficients of the acrylamide content prediction model are obtained by fitting a multiple linear regression model with the acrylamide content of the roasted coffee sample as the dependent variable and the average spectral reflectance of multiple feature bands of the roasted coffee sample as the independent variable.
[0069] Based on the selected bands, the prediction results were further optimized using MLR. The prediction results of the MLR model built using five selected bands (888, 1123, 1456, 1636, and 1734 nm) are as follows: Figure 6 As shown. The prediction results for SPA-MLR are obtained, R... p 2 The concentration was 0.81, and the RMSEP was 35.90 μg / kg. Then, the distribution and visualization of acrylamide in roasted coffee were studied using the SPA-MLR equation. This equation has good predictive performance. The test data were input into the acrylamide prediction model of this invention to obtain the acrylamide content estimation results. The SPA-MLR regression equation is:
[0070] Y = 2852.934λ 888nm +534.033λ 1123nm -5207.021λ 1456nm +5438.302λ 1636nm -1121.416λ 1734nm -719.554.
[0071] Where, λ 888nm , λ 1123nm , λ 1456nm , λ 1636nm , λ 1734nm These represent the reflectance at the corresponding characteristic wavebands.
[0072] Step 107: Update the current pixel and return to step 105 until all pixels have been traversed to obtain the acrylamide content of the roasted coffee sample at each pixel in the hyperspectral corrected image.
[0073] Step 108: Construct an acrylamide content distribution map based on the acrylamide content of the roasted coffee sample at each pixel.
[0074] Hyperspectral imaging technology is superior to spectral technology. Based on the spectral reflectance of hyperspectral images, a predictive model for acrylamide content can be established, which can be described as Y = f(x), where x and Y represent the spectral reflectance value and the acrylamide content value, respectively. Since each pixel of a hyperspectral image corresponds to a spectral reflectance value, each pixel also has an acrylamide content value. Then, the acrylamide content values of all pixels are labeled with various colors. Finally, a visual distribution map showing the acrylamide content values in roasted coffee is generated. Figures 7-8 This allows us to see the pixel-level distribution of acrylamide content in the image.
[0075] Following step 103, the following is also included:
[0076] Step 109: Smooth the spectral reflectance curve of each pixel across the entire spectral band.
[0077] The steps preceding step 101 also include:
[0078] Step 1010: The acrylamide content of multiple roasted coffee samples was determined using liquid chromatography-tandem mass spectrometry.
[0079] The method for determining acrylamide content is as follows:
[0080] (1) Sample extraction
[0081] Each batch of ground coffee sample was mixed using a DS-1 mixer (manufactured by Shanghai Precision Instruments Co., Ltd.). 1 g of homogenized sample was placed in a 50 mL Falcon tube, and acrylamide (13C3-ACR) with all carbon atoms labeled with carbon-13 isotopes (200 μg / kg, 1 mg / L internal standard) was added, along with 5 mL of hexane. The hexane layer was discarded, and the tube was vortexed to ensure matrix dispersion. Degreasing with hexane was performed twice. Subsequently, 15 mL of liquid-grade water was added to each sample, and the mixture was shaken on a horizontal shaker for 30 min. The mixture was then centrifuged at 1500 g for 5 min, and the supernatant was collected and placed in a new tube.
[0082] The aqueous extract was further purified by filtering it using a 0.45 mm solid-phase microextraction (SPE) cartridge. A cartridge was prepared using MeOH and water, and the filtered aqueous extract was then added. The extract was passed through an adsorbent, removed, and the cartridge was rinsed with 3 mL of water. Residual acrylamide on the SPE column was then eluted with 3 mL of MeOH. The eluent was evaporated to dryness in a nitrogen stream at 40 °C in a water bath, and the residue was redissolved in 0.5 mL of MeOH. Finally, the initial effluent was filtered through a 0.22 mm syringe and analyzed by ultra-high performance liquid chromatography-tandem mass spectrometry (UPLC-MS / MS).
[0083] (2) Calibration curve
[0084] Stock solutions of 3.2 mg / L acrylamide (LACR) and 1 mg / L [13C3]-ACR were prepared using MeOH. The acrylamide stock solutions were then diluted with MeOH to obtain standard solutions of 12.5 μg / L, 25.0 μg / L, 50.0 μg / L, 100.0 μg / L, 200.0 μg / L, 400.0 μg / L, and 800.0 μg / L. Acrylamide was quantified using the internal standard method, with 200 μg / kg [13C3]-AA used as the internal standard during sample preparation.
[0085] (3)UPLC-MS / MS conditions
[0086] Quantitative analysis of acrylamide was performed using a Shimadzu Nexera UPLC system and a Thermo Fisher Q-Exactive mass spectrometer. Chromatographic separation was performed at 35 °C using a Waters AcquityUPLCHSS T 3-column (2.1 mm × 100 mm, 1.8 μm). The mobile phase consisted of 0.1% (v / v) FA in water (elution A) and 0.1% FA in acetonitrile in water (elution B), with a flow rate of 0.3 mL / min and an equal volumetric elution time of 5 min (elution A 90%, eluent B 10%). The injection volume was 2 μL.
[0087] UPLC-MS / MS was performed using positive ion electrospray ionization under the following conditions: source temperature 150 °C, solvent removal temperature 500 °C, and nebulizing gas flow rate 1000 L / h. Using the selected ion monitoring mode, acrylamide showed a significant peak at 1.13 min. Furthermore, acrylamide was analyzed using 75.20 > 58.15 as an internal standard (13C3-ACR). After analysis, the sample was washed with acetonitrile-water containing 0.1% FA at a rate of 300 μL / min for 30 min.
[0088] (4) Acrylamide content determination
[0089] Acrylamide content was analyzed in six types of roasted coffee (25 samples of each). The acrylamide contents were as follows: Arabica (medium roast: 241.44–328.79 μg / kg; dark roast: 115.90–200.16 μg / kg), Catimor (medium roast: 93.10–319.25 μg / kg; dark roast: 204.58–315.27 μg / kg), and Robusta (medium roast: 361.80–533.23 μg / kg; dark roast: 204.75–282.91 μg / kg). Significant differences in acrylamide content were observed between roasted coffees from different origins and processing methods. Most acrylamide levels were below the European Commission's benchmark level for roasted coffee (400 μg / kg); however, the acrylamide content in the medium roast Robusta exceeded the standard. Because an anomalous sample (93.10 μg / kg) was found in the medium-roasted Catimor, this sample will not be considered in subsequent studies.
[0090] Step 1011: Obtain the original hyperspectral images of multiple roasted coffee samples as the original hyperspectral images of the samples.
[0091] Step 1012: Using standard whiteboard images and standard blackboard images, correct each sample's original hyperspectral image to obtain multiple corrected hyperspectral images of the samples.
[0092] Step 1013: Determine the average spectral reflectance curve of each sample hyperspectral corrected image; the average spectral reflectance curve is the average value of the spectral reflectance curves of all pixels in the corresponding sample hyperspectral corrected image across the entire spectral band.
[0093] Step 1014: Based on the average spectral reflectance curve, determine the average spectral reflectance of multiple characteristic bands corresponding to each roasted coffee sample.
[0094] Step 1015: Using the acrylamide content of the roasted coffee sample as the dependent variable and the average spectral reflectance of multiple characteristic bands corresponding to the roasted coffee sample as the independent variable, fit the multiple linear regression model to obtain the acrylamide content prediction model.
[0095] Following step 1013, the following is also included:
[0096] Step 1016: Smooth the average spectral reflectance curve.
[0097] Example 2
[0098] In order to perform the method corresponding to Embodiment 1 above and achieve the corresponding functions and technical effects, a system for determining the acrylamide content distribution in roasted coffee is provided below, comprising:
[0099] The image acquisition module is used to acquire standard whiteboard images, standard blackboard images, and the original hyperspectral image of the roasted coffee to be tested.
[0100] The hyperspectral correction image acquisition module is used to correct the original hyperspectral image of the coffee to be tested using standard whiteboard and standard blackboard images, so as to obtain the hyperspectral correction image of the coffee to be tested.
[0101] The spectral reflectance curve determination module is used to determine the full-spectral reflectance curve of each pixel in the hyperspectral corrected image of the roasted coffee to be tested.
[0102] The current pixel determination module is used to determine any pixel as the current pixel.
[0103] The feature band spectral reflectance determination module is used to obtain the spectral reflectance of multiple feature bands from the spectral reflectance curve of the current pixel across the entire spectral band.
[0104] The acrylamide content determination module is used to input the spectral reflectance of multiple characteristic bands of the current pixel into the acrylamide content prediction model to obtain the acrylamide content of the roasted coffee to be tested at the current pixel. The coefficients of the acrylamide content prediction model are obtained by fitting a multiple linear regression model with the acrylamide content of the roasted coffee sample as the dependent variable and the average spectral reflectance of multiple characteristic bands of the roasted coffee sample as the independent variable.
[0105] The multi-pixel acrylamide content determination module is used to update the current pixel and call the characteristic band spectral reflectance determination module until all pixels are traversed, so as to obtain the acrylamide content of the roasted coffee sample at each pixel in the hyperspectral corrected image.
[0106] The acrylamide content distribution determination module is used to construct an acrylamide content distribution map based on the acrylamide content of the roasted coffee sample at each pixel.
[0107] Example 3
[0108] This embodiment provides an electronic device, including a memory and a processor. The memory stores a computer program, and the processor runs the computer program to enable the electronic device to perform the method for determining the acrylamide content distribution in roasted coffee as described in Embodiment 1. The memory is a readable storage medium.
[0109] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the systems disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the descriptions are relatively simple; relevant parts can be referred to the method section.
[0110] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.
Claims
1. A method for determining the acrylamide content distribution in roasted coffee, characterized in that, include: Acquire standard whiteboard images, standard blackboard images, and the raw hyperspectral images of the coffee to be tested; Using the standard whiteboard image and the standard blackboard image, the original hyperspectral image of the coffee to be tested is corrected to obtain the corrected hyperspectral image of the coffee to be tested. Determine the full-spectral reflectance curve of each pixel in the hyperspectral corrected image of the coffee to be tested; Determine any pixel as the current pixel; The spectral reflectance of multiple feature bands is obtained from the spectral reflectance curve of the current pixel across the entire spectral band. The spectral reflectance of multiple feature bands of the current pixel is input into the acrylamide content prediction model to obtain the acrylamide content of the roasted coffee to be tested at the current pixel. The coefficients of the acrylamide content prediction model are obtained by fitting a multiple linear regression model with the acrylamide content of the roasted coffee sample as the dependent variable and the average spectral reflectance of multiple feature bands of the roasted coffee sample as the independent variable. Update the current pixel and return to the step "Obtain the spectral reflectance of multiple feature bands from the spectral reflectance curve of the full spectrum of the current pixel" until all pixels are traversed to obtain the acrylamide content of the roasted coffee sample at each pixel in the hyperspectral corrected image. A distribution map of acrylamide content is constructed based on the acrylamide content of the roasted coffee sample at each pixel.
2. The method for determining the acrylamide content distribution in roasted coffee according to claim 1, characterized in that, The standard whiteboard image has a spectral reflectance of 100%; the standard blackboard image has a spectral reflectance of 0.
3. The method for determining the acrylamide content distribution in roasted coffee according to claim 1, characterized in that, The hyperspectral corrected image is: Among them, I calibrated For hyperspectral corrected images; I raw This is the original hyperspectral image; I white A standard whiteboard image; I dark Standard blackboard image.
4. The method for determining the acrylamide content distribution in roasted coffee according to claim 1, characterized in that, After determining the spectral reflectance curve of each pixel in the hyperspectral corrected image of the coffee to be tested, the following steps are also included: The spectral reflectance curve of each pixel across the entire spectral band is smoothed.
5. The method for determining the acrylamide content distribution in roasted coffee according to claim 1, characterized in that, Before acquiring standard whiteboard images, standard blackboard images, and the original hyperspectral image of the coffee to be tested, the following steps are also included: The acrylamide content in multiple roasted coffee samples was determined using liquid chromatography-tandem mass spectrometry. Multiple roasted coffee samples were acquired as raw hyperspectral images of the samples. Using the standard whiteboard image and the standard blackboard image, each sample's original hyperspectral image is corrected to obtain multiple corrected hyperspectral images of the samples; Determine the average spectral reflectance curve of each sample hyperspectral corrected image; the average spectral reflectance curve is the average value of the spectral reflectance curves of all pixels in the corresponding sample hyperspectral corrected image across the entire spectral band. Based on the average spectral reflectance curve, the average spectral reflectance of multiple characteristic bands corresponding to each roasted coffee sample is determined. Using the acrylamide content of roasted coffee samples as the dependent variable and the average spectral reflectance of multiple characteristic bands corresponding to the roasted coffee samples as the independent variable, a multiple linear regression model was fitted to obtain an acrylamide content prediction model.
6. The method for determining the acrylamide content distribution in roasted coffee according to claim 5, characterized in that, After determining the average spectral reflectance curve of each sample's hyperspectral corrected image, the following steps are also included: The average spectral reflectance curve is smoothed.
7. A system for determining the acrylamide content distribution in roasted coffee, characterized in that, include: The image acquisition module is used to acquire standard whiteboard images, standard blackboard images, and the original hyperspectral image of the roasted coffee to be tested. The hyperspectral correction image acquisition module is used to correct the original hyperspectral image of the roasted coffee to be tested using the standard whiteboard image and the standard blackboard image, so as to obtain the hyperspectral correction image of the roasted coffee to be tested. The spectral reflectance curve determination module is used to determine the full-spectral reflectance curve of each pixel in the hyperspectral correction image of the roasted coffee to be tested. The current pixel determination module is used to determine any pixel as the current pixel. The feature band spectral reflectance determination module is used to obtain the spectral reflectance of multiple feature bands from the spectral reflectance curve of the current pixel across the entire spectral band. The acrylamide content determination module is used to input the spectral reflectance of multiple characteristic bands of the current pixel into the acrylamide content prediction model to obtain the acrylamide content of the roasted coffee to be tested at the current pixel; the coefficients of the acrylamide content prediction model are obtained by fitting a multiple linear regression model with the acrylamide content of the roasted coffee sample as the dependent variable and the average spectral reflectance of multiple characteristic bands of the roasted coffee sample as the independent variable. The multi-pixel acrylamide content determination module is used to update the current pixel and call the characteristic band spectral reflectance determination module until all pixels are traversed to obtain the acrylamide content of the roasted coffee sample at each pixel in the hyperspectral corrected image. The acrylamide content distribution determination module is used to construct an acrylamide content distribution map based on the acrylamide content of the roasted coffee sample at each pixel.
8. An electronic device, characterized in that, The device includes a memory and a processor, the memory being used to store a computer program, and the processor running the computer program to cause the electronic device to perform the method for determining the acrylamide content distribution in roasted coffee according to any one of claims 1 to 6.
9. An electronic device according to claim 8, characterized in that, The memory is a readable storage medium.