Olfactory imaging method for detecting freshness of meat based on color-sensitive sensing array

Abstract

The invention discloses a color-sensitive sensing array for meat freshness detection, the color-sensitive sensing array contains four color-sensitive materials capable of generating color development reaction with meat volatile organic compounds (VOCs), and the invention also discloses a smell imaging method for detecting meat freshness based on the color-sensitive sensing array. According to the method, the VOCs are subjected to sensitive detection through the color-sensitive sensing array, change information of the VOCs generated in a meat sample decay process is selectively extracted, and a multivariable prediction model is established by combining image processing and partial least squares regression so as to judge the freshness of meat. According to the invention, rapid and nondestructive detection of meat freshness is realized, the accuracy and efficiency of meat freshness detection can be improved, and the method is especially suitable for on-site real-time detection.

Description

Technical Field

[0001] The invention belongs to the technical field of food detection, and in particular relates to an olfactory imaging method for detecting meat freshness based on a color-sensitive sensor array. Background Art

[0002] Food safety is a matter of great importance. Meat, as an important part of human diet, is rich in nutrients such as protein, fat, iron, zinc and multiple vitamins, and is one of the indispensable foods for maintaining health. With the growth of global population and economy, the demand for meat products has increased year by year. Survey data shows that the average daily meat intake per capita in China is 89.7 grams. However, during the transportation and storage of meat, due to temperature fluctuations or improper environment, fat is easily oxidized and the surface is easily contaminated by microorganisms, resulting in a decrease in freshness and corruption. Therefore, it is particularly necessary to quickly and accurately detect the freshness of meat.

[0003] Volatile basic nitrogen (TVB-N) is produced during the meat spoilage process due to the decomposition of proteins by enzymes and bacteria, producing alkaline nitrogen-containing substances such as ammonia and amines. These substances are volatile, and the higher the TVB-N content, the more protein is destroyed and the more severe the loss of freshness. TVB-N is one of the commonly used indicators to measure the freshness of meat. In addition, meat spoilage is also accompanied by the production of volatile organic compounds (VOCs), such metabolites include low molecular weight aldehydes, alcohols and ketones. The content and composition of VOCs are closely related to the meat spoilage process and can be used as a biomarker for meat freshness.

[0004] Traditional meat freshness detection methods such as gas chromatography-mass spectrometry (GC-MS) are highly accurate, but the equipment is expensive, the operation is complex, time-consuming and requires professional personnel to operate, making it difficult to use for on-site real-time detection. In contrast, olfactory imaging technology based on color-sensitive materials reacts with specific gases to produce color changes, combined with digital image processing to quantitatively or qualitatively analyze volatile gases, and has become an effective means of quickly detecting the freshness of food. Commonly used color-sensitive materials include pH indicators, porphyrins and fluoroboryl pyrrole compounds. Among them, fluoroboryl pyrrole compounds have a high responsiveness to volatile gases such as aldehydes and ketones, but there is cross-reactivity, and the accuracy and sensitivity are limited.

[0005] Existing olfactory imaging technology has problems such as complex operation, long time consumption and low detection accuracy, which makes it difficult to meet the real-time monitoring needs of meat in transportation and storage. Based on this, the present invention has developed a color-sensitive sensor array that can achieve high-sensitivity monitoring of volatile gases in meat, and then achieve continuous detection through image acquisition training modeling combined with certain algorithms, which is suitable for rapid and non-destructive detection of meat freshness. Summary of the invention

[0006] The first object of the present invention is to provide a color-sensitive sensor array for detecting meat freshness, and the second object of the present invention is to provide an olfactory imaging method for detecting meat freshness based on the color-sensitive sensor array. The present invention can realize rapid, continuous, and non-destructive detection of meat freshness, and also has the characteristics of simple operation, low cost, and high detection sensitivity.

[0007] To achieve the above object, the technical solution of the present invention is as follows: A color-sensitive sensor array for detecting meat freshness contains four color-sensitive materials that can produce color reactions with volatile organic compounds (VOCs) in meat. The four color-sensitive materials are meso-tetra(4-sulfophenyl)porphine tetrasodium dodecahydrate, 8-(4-nitrophenyl)fluoroborylpyrrole, 8-(2,4,6-trimethylphenyl)-1,3,5,7-tetramethyl-fluoroboryl dipyrrole, and dimethyl[(1-pyrrolidinyl)phenyl]-amine nickel(II) chloride.

[0008] The preparation method of the color-sensitive sensor array is as follows: the four color-sensitive materials are dissolved in organic solvents respectively, and then coated on a substrate material to form four sensor arrays.

[0009] Wherein, the base material is a polyvinylidene fluoride film.

[0010] The color-sensitive sensor array provided by the present invention can produce a color reaction with VOCs produced by meat. The four color-sensitive materials used have high responsiveness and strong selectivity in reacting with VOCs. The content of VOCs is analyzed by the degree of color change, and the freshness of the meat is then judged. Therefore, it has a good application prospect in meat freshness detection.

[0011] The present invention further provides an olfactory imaging method for detecting meat freshness based on a color-sensitive sensor array, comprising the following steps: S1. Obtain modeling samples of different freshness; S2, placing the modeling sample and the color-sensitive sensor array in a sealed container for reaction, and using a digital camera to collect images of the color-sensitive sensor array before and after the reaction; S3, processing the collected images, and then calculating the sensor array characteristic data of each color-sensitive material in three color spaces: RGB, HSV, and LAB; S4. Detect the TVB-N value of the modeling sample; S5. Based on the sensor array characteristic data and TVB-N value, a meat freshness prediction model is constructed using partial least squares regression; S6. Use the model to predict the TVB-N value of the meat sample and determine the freshness of the sample based on the TVB-N value.

[0012] Wherein, in step S1, the meat sample is stored at 4°C, and samples are taken regularly to obtain modeling samples of different freshness.

[0013] Wherein, in step S2, the reaction temperature is 20-30°C.

[0014] Among them, in step S3, the captured image is first subjected to threshold segmentation, edge detection, perspective transformation, and feature area recognition, and then interference factors, corrosion, and expansion are removed according to the morphological processing method, and finally the sensor array feature data in the three color spaces of RGB, HSV, and LAB are extracted according to the contour.

[0015] The beneficial effects of the present invention are: (1) High sensitivity and selectivity: The color-sensitive sensing material can accurately detect the characteristic volatile gases produced during meat spoilage; (2) Rapid and non-destructive testing: It can realize continuous and real-time meat freshness testing without damaging the meat samples during the testing process; (3) Real-time online monitoring: Through the combination of image acquisition system and data processing module, real-time online monitoring of meat freshness is realized, which is suitable for on-site use; (4) Applicable to a variety of meats: It can be widely used in different types of meat, such as pork, beef and poultry, and has strong versatility; (5) Intelligence and automation: Use machine learning algorithms to analyze and predict the freshness of meat, with a high degree of automation and reduced human intervention. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 The PCA diagram of the four screened color-sensitive materials.

[0017] Figure 2 Schematic diagram of the color-sensitive sensor array.

[0018] Figure 3 This is the flow chart of feature image extraction.

[0019] Figure 4 This is the TVB-N variation trend chart of pork samples.

[0020] Figure 5 Performance analysis results of the PLS-based pork TVB-N prediction model. DETAILED DESCRIPTION

[0021] The technical solution of the present invention will be further explained below in conjunction with specific embodiments.

[0022] Example 1: Preparation of sensor array and detection of pork freshness 1. Screening of color-sensitive materials In terms of material selection, this study conducted experiments on 15 color-sensitive materials by consulting the literature and screened out the most suitable materials for detecting the freshness of meat. These 15 materials include 5 porphyrins and 10 pyrrole color-sensitive materials (specific types and numbers are as follows). Experimental process: The first sampling was carried out immediately after the pork samples (including tenderloin, rump meat, belly meat and hind leg meat) were purchased on the first day, recorded as 0 hours, and the samples were stored at 4°C. Then, samples were sampled every 24 hours for 4 days. In order to ensure the randomness and reliability of the samples, 10g of samples were randomly collected each time the pork parts were not distinguished, and a total of 40 sample data were obtained. These data helped us record the changes in the freshness of pork within 4 days.

[0023] Porphyrin compounds: tetraphenylporphyrin (T1), tetraphenylporphyrin cobalt (T8), tetraphenylporphyrin zinc (T9), meso-tetrakis (4-sulfophenyl) porphine tetrasodium dodecahydrate (T14), tetraphenylporphyrin nickel (T19).

[0024] Pyrrole compounds: 8-(4-carbazolephenyl)-4,4-difluoroborylpyrrolmethane (S1), 8-(4-nitrophenyl)-4,4-difluoro-2,6-dibromoborylpyrrolmethane (S2), 8-(4-nitrophenyl)fluoroborylpyrrole (S3), 8-(2,4,6-trimethylphenyl)-1,3,5,7-tetramethyl-fluoroborylpyrrole (S5), 8-phenyl-4,4-difluoroborylpyrrolmethane (S6), 8 -(6-methoxy-2-naphthyl)-4,4-difluoroborondipyrromethane (S8), bis[8-phenyldipyrromethane]nickel(II) (S11), dimethyl[(1-pyrrolidinyl)phenyl]-aminenickel(II) chloride (S12), bis[8-(4-methylformylphenyl)dipyrromethane]nickel(II) (S13), bis[8-(6-methoxy-2-naphthyl)dipyrromethane]nickel(II) (S14).

[0025] When screening for color-sensitive materials, a two-step approach was used: In the first step, in the RGB color space, the mean and standard deviation of the R, G, and B components of each color-sensitive material are calculated, and the ratio of the mean to the standard deviation is used as the screening criterion. If the ratio is lower than 2:1, it means that the response fluctuation of the material is large and it is not suitable for freshness detection. Specifically (see Table 1), the black bold part indicates that the material does not meet the standard, while the ratios of materials T14, S3, S5, and S12 meet the standard and are selected for further analysis.

[0026] Table 1 Mean and deviation statistics of 15 color-sensitive materials in RGB color space

[0027] In the second step, the principal component analysis (PCA) method was used to conduct a deeper data analysis on the selected materials. Its function is to detect whether the response of these materials under different freshness is stable and has sufficient discrimination ability. The PCA model was constructed by MATLAB 2014b software to detect the stability and discrimination effect of the materials on the change of freshness. The analysis results show (such as Figure 1 As shown in the figure, the four color-sensitive materials T14, S3, S5 and S12 have a high differentiation effect at different storage stages and can accurately reflect the changes in pork freshness. Based on this, it is confirmed that these four materials are suitable for pork freshness detection and have high application value.

[0028] 2. Preparation of Sensor Array In this study, four materials sensitive to volatile organic compounds (VOCs) were finally selected, namely meso-tetra(4-sulfophenyl)porphine tetrasodium dodecahydrate (T14), 8-(4-nitrophenyl) fluoroborylpyrrole (S3), 8-(2,4,6-trimethylphenyl)-1,3,5,7-tetramethyl-fluoroboryl dipyrrole (S5), and dimethyl[(1-pyrrolidinyl)phenyl]-amine nickel(II) chloride (S12). These materials were selected because they have high sensitivity to VOCs and can effectively distinguish the state of pork at different freshness stages.

[0029] During the sensor preparation process, S3, S5, and S12 were dissolved in formamide solvent, while T14 was dissolved in ethanol, and the concentration of all solutions was 2 mg / mL. Then, these solutions were evenly drop-coated on a polyvinylidene fluoride (PVDF) membrane to form a multi-color sensitive sensor array (see Figure 2 ). When pork samples release VOCs, this sensor array can directly display the changes in VOCs concentration through color changes, thereby making an intuitive judgment on the freshness of pork and being able to sensitively detect changes in the freshness of pork.

[0030] 3. Construction of pork freshness detection model (1) Preparation of pork samples: Different parts of pork were selected, including tenderloin, rump, belly, and hind leg, and stored at 4°C for 96 hours. To ensure the randomness and reliability of the samples, 20 samples of pork were randomly selected from different parts of the pork for testing every 12 hours. A total of 180 samples were obtained, and each sample was cut into 10 g pieces for testing.

[0031] (2) Detection of volatile gases: Place the pork sample at the bottom of a sealed container and place the sensor array above the container. Allow the reaction to take place at 26°C for 10 minutes, and collect image data of the sensor array before and after the reaction. Remove the original image from the post-reaction image to obtain the characteristic data of the sensor array for volatile gases in pork.

[0032] (3) Image acquisition and feature extraction of color-sensitive sensor array: The image data before and after the color-sensitive sensor array reaction is collected by 3CCD industrial camera, and Figure 3 The image extraction algorithm shown in the figure starts from threshold segmentation, edge detection, perspective transformation, feature area recognition, and then dilation and corrosion are used to remove interference factors according to the morphological processing method. Finally, the color sensitive sensor array is extracted according to the contour, and the mean of the feature data of each color sensitive material is calculated in the three color spaces of RGB, HSV, and LAB. Each sample generates 36 variables. Taking one of the samples as an example, the specific values ​​of the 36 variables are shown in Table 2.

[0033] Table 2 Data style tables in various color spaces

[0034] (4) Pork freshness test: After the volatile gas test, the volatile basic nitrogen (TVB-N) content of pork is measured using a fully automatic Kjeldahl nitrogen analyzer. The results are as follows: Figure 4 As shown. During storage, the TVB-N content gradually increased, and the difference between samples in the same batch was less than 0.3 mg / 100 g. According to national standards, the boundary between freshness and staleness of pork is 15 mg / 100 g. After reaching the critical point, the TVB-N content increases rapidly.

[0035] (5) Establish a pork freshness prediction model using sensor array data and TVB-N value: A pork freshness prediction model was constructed based on partial least squares regression (PLS). The independent variable is the feature data extracted after the color-sensitive sensor array detects the volatile gas of pork, including 36 variables in three color spaces: RGB, HSV, and LAB. The dependent variable is the content of pork volatile basic nitrogen (TVB-N), which reflects the freshness of pork. The samples were divided into training set and test set in a ratio of 3:2, and the corresponding PLS function library was called using MATLAB2014b software for regression modeling.

[0036] During the modeling process, the data was standardized to prevent variables of different dimensions from having too much impact on the model. The standardization formula is as follows: Standardization formula for independent variable X: Standardization formula of dependent variable Y:

[0037] in, and are the means of the independent and dependent variables, respectively. and are the standard deviations of the independent and dependent variables, respectively.

[0038] The optimal number of latent variables was determined to be 8 through cross-validation ( Figure 5 Left), these latent variables extract the most relevant information about the target variable through feature extraction and data dimensionality reduction, solve the multicollinearity problem and remove noise, improve the stability and predictive ability of the model. The model performance is measured by the correlation coefficient (R C ), prediction correlation coefficient (R P ), cross validation root mean square error (RMSEC) and prediction set root mean square error (RMSEP) for evaluation. Figure 5 As shown on the right, R C = 0.9846, R P =0.9835 indicates that the model has high fitting and prediction capabilities, RMSEC = 0.4662, RMSEP = 0.4988 indicates that the model error is small and has good generalization ability. The study shows that the PLS model effectively captures the relationship between sensor array data and TVB-N content and can effectively predict the freshness of pork.

[0039] 4. Verification of the actual detection effect of the prediction model Ten pork samples (regardless of the part) were randomly selected for testing, and the results are shown in Table 3. According to the TVB-N values, eight of the samples were judged to be fresh pork and two were judged to be stale pork. The TVB-N values ​​predicted by the model were generally close to the actual test values. From the final prediction results, the prediction model only made a wrong judgment on one sample (the bold black part), and judged fresh pork as stale pork, with a correct judgment rate of 90%. This shows that the constructed PLS prediction model has an ideal effect.

[0040] Table 3 PLS-based model measurement results

[0041] In summary, this study developed a specific color-sensitive sensor array for pork volatile gases based on olfactory imaging technology, and established a corresponding prediction model. The research results show that the model has a stable and efficient prediction effect. The present invention is universal for the detection of meat foods. When using the color-sensitive sensor array to detect other meat foods such as beef, mutton, chicken, duck, fish, etc., it is only necessary to re-model according to the sensor array data and TVB-N value.

Claims

1. A color-sensitive sensor array for detecting meat freshness, comprising four color-sensitive materials that can produce colorimetric reactions with meat volatile organic compounds (VOCs), the four color-sensitive materials being meso-tetra(4-sulfophenyl)porphine tetrasodium dodecahydrate, 8-(4-nitrophenyl)fluoroborylpyrrole, 8-(2,4,6-trimethylphenyl)-1,3,5,7-tetramethyl-fluoroboryl dipyrrole, and dimethyl[(1-pyrrolidinyl)phenyl]-amine nickel(II) chloride.

2. A method for preparing the color-sensitive sensor array according to claim 1, characterized in that: The four color-sensitive materials are dissolved in organic solvents respectively, and then coated on a substrate material to form four sensing arrays.

3. The method for preparing a color sensitive sensor array according to claim 2, characterized in that: The substrate material is a polyvinylidene fluoride film.

4. Application of the color-sensitive sensor array described in claim 1 in detecting meat freshness.

5. An olfactory imaging method for detecting meat freshness based on a color-sensitive sensor array, characterized in that The following steps are involved: S1. Obtain modeling samples of different freshness; S2, placing the modeling sample and the color-sensitive sensor array according to claim 1 in a sealed container for reaction, and collecting images of the color-sensitive sensor array before and after the reaction with a digital camera; S3, processing the collected images, and then calculating the sensor array characteristic data of each color-sensitive material in three color spaces: RGB, HSV, and LAB; S4. Detect the TVB-N value of the modeling sample; S5. Based on the sensor array characteristic data and TVB-N value, a meat freshness prediction model is constructed using partial least squares regression; S6. Use the model to predict the TVB-N value of the meat sample and determine the freshness of the sample based on the TVB-N value.

6. The olfactory imaging method for detecting meat freshness according to claim 5, characterized in that: In step S1, the meat samples are stored at 4°C and samples are taken regularly to obtain modeling samples with different freshness.

7. The olfactory imaging method for detecting meat freshness according to claim 5, characterized in that: In step S2, the reaction temperature is 20-30°C.

8. The olfactory imaging method for detecting meat freshness according to claim 5, characterized in that: In step S3, the captured image is first subjected to threshold segmentation, edge detection, perspective transformation, and feature region recognition, and then interference factors, corrosion, and expansion are removed according to the morphological processing method, and finally the sensor array feature data in the three color spaces of RGB, HSV, and LAB are extracted according to the contour.

Images