Information processing method based on food detection

Abstract

The invention relates to the technical field of food detection, and discloses an information processing method based on food detection. According to the method, food components are analyzed through a spectrograph and a chromatographic instrument, data and images of food analysis are recorded, missing values, abnormal values and repeated data are removed from the recorded food data through an algorithm formula, the data of different dimensions are converted into a unified standard, and the data are normalized to eliminate the influence of the dimensions. Extracting a color histogram and texture features from the processed food data, selecting features which have the greatest influence on the model by applying a principal component analysis method, selecting a support vector machine algorithm according to data features, performing model training by using a labeled data set, adjusting parameter optimization performance, and evaluating the generalization ability of the model by using a cross validation method. Evaluating the performance and classification effect of the model, detecting food through the trained model, summarizing detection results, generating a detection analysis report, and outputting the detection analysis report to a display end.

Description

Technical Field

[0001] The present invention relates to the technical field of food detection, and in particular to an information processing method based on food detection. Background Art

[0002] Food safety refers to the ability of food to maintain its quality and safety during the process of production, processing, storage and consumption, ensuring that it does not cause harm to consumers' health. It is directly related to consumers' health. In the context of globalization, food safety standards have become an important part of international trade. Therefore, food testing is an important means to ensure food safety. Through testing, harmful substances in food such as pesticide residues, heavy metals and additives can be identified to ensure that food meets safety standards. At the same time, regular food testing can detect potential safety hazards early and prevent the occurrence of food safety incidents.

[0003] Although food testing plays an important role in ensuring food safety, there are still some shortcomings in information processing. Data between different testing agencies and laboratories often lack unified standards, resulting in information islands, making it difficult to effectively integrate and share data, and it is difficult to extract valuable information from the data in terms of analysis and processing capabilities of test data. Summary of the invention

[0004] 1. Technical issues to be solved

[0005] In view of the deficiencies in the prior art, the present invention provides an information processing method based on food detection, which has the functions of performing food component analysis through a spectrometer and a chromatograph, measuring the pH value, sugar, salt and fat content of food, recording food analysis data and images, removing missing values, outliers and duplicate data from the above-mentioned recorded food data using an algorithm formula, converting data of different dimensions into a unified standard, and normalizing the data to eliminate the influence of the dimension, extracting color histogram and texture features from the processed food data, applying the principal component analysis method to select the features that have the greatest impact on the model, selecting a support vector machine algorithm according to data characteristics, using annotated data sets to train the model, adjusting parameters to optimize performance, and using a cross-validation method to evaluate the generalization ability of the model, evaluating the performance and classification effect of the model, detecting food through a trained model, summarizing the detection results to generate a detection analysis report and outputting it on a display end, and solving the above-mentioned problems by standardizing food data, effectively integrating and sharing data, avoiding the phenomenon of information islands, and being able to extract valuable information from the detected food data.

[0006] (II) Technical solution

[0007] To achieve the above object, the present invention provides the following technical solution: an information processing method based on food detection, comprising the following steps:

[0008] S1. Obtain food samples, analyze food ingredients using a spectrometer and chromatograph, measure food pH, sugar, salt and fat content, and record food analysis data and images;

[0009] S2. Remove missing values, outliers and duplicate data from the above-mentioned recorded food data through an algorithm formula, convert data of different dimensions into a unified standard, and normalize the data to eliminate the impact of dimensions;

[0010] S3, extracting color histogram and texture features from the processed food data, and applying principal component analysis to select the features that have the greatest impact on the model;

[0011] S4. Select the support vector machine algorithm according to the data characteristics, use the labeled data set to train the model, adjust the parameters to optimize the performance, and use the cross-validation method to evaluate the generalization ability of the model;

[0012] S5. Evaluate the performance of the model, calculate the model accuracy Mxzq, model recall rate Mxzh, model F1 score Mxfs, and evaluate the classification effect of the model;

[0013] S6. Test the food using the trained model, summarize the test results, generate a test analysis report and output it to the display terminal.

[0014] Preferably, S2 removes missing values ​​from the food data recorded above through an algorithm formula, and the formula for removing missing values ​​is as follows:

[0015]

[0016] In the formula, Wzsj represents the food dataset with missing values ​​removed, Spsj represents the original food dataset, and Spsj i represents the i-th original food sample, x j represents the jth food feature.

[0017] Preferably, S2 removes outliers from the recorded food data using an algorithm formula, and the outlier removal formula is as follows:

[0018]

[0019] Qcqs={Spsj i ∈Wzsj||Zxsd n | <thod}

[0020] In the formula, Zxsd nrepresents the Z-score of the i-th food sample, X n represents the food feature value of the nth sample, μ represents the mean of the feature, σ represents the standard deviation of the feature, Z-score measures the degree of deviation of the sample value from the mean, and the unit is standard deviation, Qcqs represents the data set with outliers removed, Spsj i represents the i-th original food sample, Wzsj represents the food data set with missing values ​​removed, and thod takes a value of 3, indicating that samples within three times the standard deviation of the mean are considered normal values. This formula means that samples with Z-score absolute values ​​less than the threshold are screened out from the data set after removing missing values ​​to form a new data set.

[0021] Preferably, S2 removes duplicate data from the recorded food data using an algorithm formula, and the formula for removing duplicate data is as follows:

[0022]

[0023] In the formula, Qccf represents the food dataset with duplicate data removed, Qcqs k represents the kth food data after removing outliers, Qcqs represents the data set after removing outliers, Qcqs m represents the mth food data after removing outliers, P i , P j Represents the feature vectors of the i-th and j-th food samples. This formula means that samples that do not have the same feature vectors as other samples are screened out from the data set after removing outliers to form a new data set.

[0024] Preferably, S2 removes missing values, outliers and duplicate data from the food data recorded above through an algorithm formula, and converts data of different dimensions into a unified standard. The conversion formula is as follows:

[0025]

[0026] In the formula, Gy represents the normalized eigenvalue, Ys represents the original eigenvalue, and Ys min 、Ys max Represents the minimum and maximum values ​​of the feature. This formula means scaling the feature value to the range of [0,1] so that data of different dimensions can be compared under the same standard.

[0027] Preferably, S3 extracts a color histogram from the processed food data, and the extraction formula is as follows:

[0028]

[0029] In the formula, H(sp) represents the histogram of color sp, I(x,y) represents the color value of image I at position (x,y), δ(sp,I(x,y)) represents the Kronecker delta function, which takes the value of 1 when I(x,y)=sp, otherwise it takes the value of 0. This formula means traversing each pixel in the image and counting the number of times each color appears.

[0030] Preferably, S3 extracts texture features from the processed food data, and the extraction formula is as follows:

[0031]

[0032] In the formula, Cons represents the texture characteristics of the food image, P(i,j) represents the co-occurrence frequency of gray level i and gray level j, and contrast measures the degree of change of gray levels in the image. The larger the value, the more complex the texture.

[0033] Preferably, S5 evaluates the performance of the model and calculates the model accuracy Mxzq, and the calculation formula is as follows:

[0034]

[0035] In the formula, Mxzq represents the model accuracy, TP represents the number of food samples that are actually positive and correctly classified as positive, TN represents the number of food samples that are actually negative and correctly classified as negative, FP represents the number of food samples that are actually negative but mistakenly classified as positive, and FN represents the number of food samples that are actually positive but mistakenly classified as negative.

[0036] Preferably, S5 evaluates the performance of the model and calculates the model recall rate Mxzh, and the calculation formula is as follows:

[0037]

[0038] In the formula, Mxzh represents the model recall rate, TP represents the number of food samples that are actually positive and correctly classified as positive, and FN represents the number of food samples that are actually positive but misclassified as negative.

[0039] Preferably, the S5 evaluates the performance of the model and calculates the model F1 score Mxfs, and the calculation formula is as follows:

[0040]

[0041] In the formula, Mxfs represents the model F1 score, Mxzq represents the model accuracy, and Mxzh represents the model recall rate.

[0042] Compared with the prior art, the present invention provides an information processing method based on food detection, which has the following beneficial effects:

[0043] The present invention uses a spectrometer and a chromatograph to analyze food ingredients, measures food pH, sugar, salt and fat content, records food analysis data and images, uses an algorithm formula to remove missing values, outliers and duplicate data from the recorded food data, converts data of different dimensions into a unified standard, and normalizes the data to eliminate the impact of the dimension, extracts color histogram and texture features from the processed food data, uses a principal component analysis method to select the features that have the greatest impact on the model, selects a support vector machine algorithm according to data characteristics, uses annotated data sets to train the model, adjusts parameters to optimize performance, and uses a cross-validation method to evaluate the generalization ability of the model, evaluates the performance and classification effect of the model, detects food through the trained model, summarizes the detection results to generate a detection analysis report and outputs it to a display end, and by standardizing and unifying food data, effectively integrates and shares data, avoids information islands, and can extract valuable information from the detected food data. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 It is a schematic diagram of the steps of the method of the present invention. DETAILED DESCRIPTION

[0045] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0046] In view of the current problem of food information processing, there is a lack of unified standards for data between different testing institutions and laboratories, which leads to information islands, making it difficult to effectively integrate and share data, and to analyze and process test data, and to extract valuable information from the data. Therefore, a method for information processing based on food testing is proposed. Figure 1 , the method comprises the following steps:

[0047] S1. Obtain food samples, analyze food ingredients using a spectrometer and chromatograph, measure food pH, sugar, salt and fat content, and record food analysis data and images;

[0048] Before conducting food component analysis, it is necessary to first obtain representative samples from the target food. This process should follow strict sampling standards to ensure the representativeness and accuracy of the samples. Samples can be obtained through random sampling, stratified sampling, etc. to ensure that the selected samples can reflect the characteristics of the entire batch of food;

[0049] Select near-infrared spectroscopy to measure the moisture, sugar and fat content in food, and UV-visible spectroscopy to analyze pigments, antioxidants and other ingredients in food;

[0050] Gas chromatography was selected for the analysis of volatile components, and liquid chromatography was used for the analysis of non-volatile components;

[0051] After completing the food composition analysis, the measured food value, sugar, salt and fat content data are organized into tables. During the analysis process, the image acquisition equipment is used to record the appearance and spectral image of the sample, and the data are input into the computer system for the next step of analysis;

[0052] S2. Remove missing values, outliers and duplicate data from the above-mentioned recorded food data through an algorithm formula, convert data of different dimensions into a unified standard, and normalize the data to eliminate the impact of dimensions;

[0053] The formula for removing missing values ​​is as follows:

[0054]

[0055] Removing missing values ​​can significantly improve the overall quality of the data set. By removing these missing values, we can ensure that the data on which the analysis is based is accurate and reliable, thereby improving the credibility of the research results. In the formula, Wzsj represents the food data set with missing values ​​removed, Spsj represents the original food data set, and Spsj i represents the i-th original food sample, x j represents the jth food feature. When performing statistical analysis or machine learning modeling, missing values ​​may cause model bias and instability. After removing missing values, the analysis results will be more accurate and can better reflect the true situation of food ingredients.

[0056] The formula for removing outliers is as follows:

[0057]

[0058] Qcqs={Spsj i ∈Wzsj||Zxsd n | <thod}

[0059] The presence of outliers may affect the data visualization effect and make the chart difficult to interpret. After removing the outliers, the generated chart will be clearer and can better show the distribution and trend of food ingredients, making it easier for researchers and stakeholders to understand the data. In the formula, Zxsd n represents the Z-score of the i-th food sample, X nrepresents the food feature value of the nth sample, μ represents the mean of the feature, σ represents the standard deviation of the feature, Z-score measures the degree of deviation of the sample value from the mean, and the unit is standard deviation, Qcqs represents the data set with outliers removed, Spsj i represents the i-th original food sample, Wzsj represents the food data set with missing values ​​removed, and thod takes a value of 3, indicating that samples within three times the standard deviation of the mean are considered normal values. This formula means that samples with Z-score absolute values ​​less than the threshold are screened out from the data set after missing values ​​are removed to form a new data set. The existence of outliers may increase the complexity of data processing, resulting in the need for additional steps to process these values. By removing outliers, the process of data cleaning and preprocessing can be simplified, reducing the time and energy investment in data analysis;

[0060] The formula for removing duplicate data is as follows:

[0061]

[0062] Duplicate data may lead to deviation and inconsistency in the analysis results. By removing duplicate data, we can ensure that each sample or observation value appears only once in the data set, thereby improving the accuracy and reliability of the data. In the formula, Qccf represents the food data set with duplicate data removed, and Qcqs k represents the kth food data after removing outliers, Qcqs represents the data set after removing outliers, Qcqs m represents the mth food data after removing outliers, P i , P j Represents the characteristic vectors of the i-th and j-th food samples. This formula indicates that samples without the same characteristic vectors as other samples are screened out from the data set after outliers are removed to form a new data set. By applying the data set with duplicate data removed, it can be ensured that the data meets the prerequisites of the analysis method and improve the accuracy of statistical tests.

[0063] The data is converted into a unified standard formula as follows:

[0064]

[0065] Data from different sources or under different experimental conditions may use different units, formats or standards. By converting food data to a unified standard, these differences can be eliminated and the consistency of the data can be ensured. In the formula, Gy represents the normalized eigenvalue, Ys represents the original eigenvalue, and Ys min 、Ys maxRepresents the minimum and maximum values ​​of the feature. This formula indicates scaling the feature value to the range of [0,1] so that data of different dimensions can be compared under the same standard. When conducting multi-source data analysis, a unified standard can help effectively integrate data from different sources, which can more comprehensively analyze food ingredients, quality, and safety, and support more in-depth research;

[0066] S3, extracting color histogram and texture features from the processed food data, and applying principal component analysis to select the features that have the greatest impact on the model;

[0067] The color histogram extraction formula is as follows:

[0068]

[0069] The color of food is an important indicator of its freshness and quality. By extracting color histogram features, the quality of food can be effectively evaluated, helping to identify spoiled or substandard products, thereby improving food safety. In the formula, H(sp) represents the histogram of color sp, I(x,y) represents the color value of image I at position (x,y), and δ(sp,I(x,y)) represents the Kroneckerdelta function. When I(x,y)=sp, the value is 1, otherwise it is 0. This formula means traversing each pixel in the image and counting the number of times each color appears. The color histogram can quantify the color information in the image and provide the frequency of each color appearing in the image. This quantification allows color features to be used for further analysis and comparison, helping researchers better understand the appearance characteristics of food.

[0070] The texture feature extraction formula is as follows:

[0071]

[0072] Texture features can quantify the details and structure of food surfaces, including roughness, smoothness, and uniformity. This quantification enables researchers to better understand the physical properties of food and thus make more accurate quality assessments. In the formula, Cons represents the texture features of food images, P(i,j) represents the co-occurrence frequency of grayscale level i and grayscale level j, and contrast measures the degree of grayscale variation in the image. The larger the value, the more complex the texture. At the same time, the texture features of food are closely related to its freshness and quality. By extracting texture features, we can effectively monitor changes in food quality and help identify spoiled or substandard products, thereby improving food safety.

[0073] S4. Select the support vector machine algorithm according to the data characteristics, use the labeled data set to train the model, adjust the parameters to optimize the performance, and use the cross-validation method to evaluate the generalization ability of the model;

[0074] Select the support vector machine algorithm according to the data characteristics, and use the labeled data set to train the model to obtain the classifier. First, select a linear kernel to adapt to the data distribution, then adjust the hyperparameters, such as the penalty parameter C and the parameters of the kernel function, to optimize the model performance. Then apply the k-fold cross-validation method to evaluate the generalization ability of the model. Through stratified sampling, ensure that the label distribution of the training set and the test set is consistent, so as to obtain more robust performance indicators, and finally ensure that the model can effectively process unseen data and improve prediction accuracy.

[0075] S5. Evaluate the performance of the model, calculate the model accuracy Mxzq, model recall rate Mxzh, model F1 score Mxfs, and evaluate the classification effect of the model;

[0076] The model accuracy calculation formula is as follows:

[0077]

[0078] The calculation of accuracy can be used as a basis for optimizing algorithms, guiding researchers to make more informed decisions when performing feature engineering, selecting different algorithms, or adjusting parameters to improve the overall model effect. In the formula, Mxzq represents the model accuracy, TP represents the number of food samples that are actually positive and correctly classified as positive, TN represents the number of food samples that are actually negative and correctly classified as negative, FP represents the number of food samples that are actually negative but incorrectly classified as positive, and FN represents the number of food samples that are actually positive but incorrectly classified as negative.

[0079] The model recall calculation formula is as follows:

[0080]

[0081] The recall rate measures the proportion of positive examples correctly identified by the model among all actual positive examples. For food safety testing tasks, it can effectively evaluate the model's ability to identify true positive samples and ensure that all qualified foods are detected as much as possible. In the formula, Mxzh represents the model recall rate, TP represents the number of food samples that are actually positive and correctly classified as positive, and FN represents the number of food samples that are actually positive but incorrectly classified as negative. In food safety management, it is very important to give priority to the detection and control of unqualified foods. Models with high recall rates can ensure that potential risks are found in a wider range of food samples and optimize decision-making and resource allocation.

[0082] The model F1 score calculation formula is as follows:

[0083]

[0084] The F1 score is the harmonic mean of precision and recall, which can comprehensively reflect the classification performance of the model, especially when there is a class imbalance, providing a more comprehensive evaluation. In the formula, Mxfs represents the model F1 score, Mxzq represents the model accuracy, and Mxzh represents the model recall. By combining precision and recall, the F1 score can seek a balance between the accuracy and completeness of the model to ensure the accuracy and safety of food testing.

[0085] S6. Detect food using the trained model, summarize the detection results, generate a detection analysis report and output it to the display terminal;

[0086] To test food through a trained model, first input the sample to be tested into the model, use feature extraction methods to obtain relevant data, and the model classifies and evaluates the samples according to the rules learned during the training process, and calculates the confidence score of each sample. After the test is completed, the results are organized into structured data, including the test category, confidence, possible risk level and recommended treatment measures. Finally, this information is summarized to generate a detailed test analysis report, which is output to the display end for users to view and make decisions. The report should include charts and visualization elements to enhance the information communication effect and help users quickly understand the test results.

[0087] The food detection model constructed by the above method can standardize food data, conduct effective data integration and sharing, avoid information island phenomenon, and extract valuable information from the detected food data.

[0088] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An information processing method based on food detection, characterized in that: The following steps are involved: S1. Obtain food samples, analyze food ingredients using a spectrometer and chromatograph, measure food pH, sugar, salt and fat content, and record food analysis data and images; S2. Remove missing values, outliers and duplicate data from the above-mentioned recorded food data through an algorithm formula, convert data of different dimensions into a unified standard, and normalize the data to eliminate the impact of dimensions; S3, extracting color histogram and texture features from the processed food data, and applying principal component analysis to select the features that have the greatest impact on the model; S4. Select the support vector machine algorithm according to the data characteristics, use the labeled data set to train the model, adjust the parameters to optimize the performance, and use the cross-validation method to evaluate the generalization ability of the model; S5. Evaluate the performance of the model, calculate the model accuracy Mxzq, model recall rate Mxzh, model F1 score Mxfs, and evaluate the classification effect of the model; S6. Test the food using the trained model, summarize the test results, generate a test analysis report and output it to the display terminal.

2. The information processing method based on food detection according to claim 1, characterized in that: S2 removes missing values ​​from the food data recorded above through an algorithm formula. The formula for removing missing values ​​is as follows: In the formula, Wzsj represents the food dataset with missing values ​​removed, Spsj represents the original food dataset, and Spsj i represents the i-th original food sample, x j represents the jth food feature.

3. The information processing method based on food detection according to claim 2 is characterized in that: S2 removes outliers from the food data recorded above through an algorithm formula. The outlier removal formula is as follows: Qcqs={Spsj i ∈Wzsj||Zxsd n |<thod} In the formula, Zxsd n represents the Z-score of the i-th food sample, X n represents the food feature value of the nth sample, μ represents the mean of the feature, σ represents the standard deviation of the feature, Z-score measures the degree of deviation of the sample value from the mean, and the unit is standard deviation, Qcqs represents the data set with outliers removed, Spsj i represents the i-th original food sample, Wzsj represents the food data set with missing values ​​removed, and thod takes a value of 3, indicating that samples within three times the standard deviation of the mean are considered normal values. This formula means that samples with Z-score absolute values ​​less than the threshold are screened out from the data set after removing missing values ​​to form a new data set.

4. The information processing method based on food detection according to claim 3 is characterized in that: S2 removes duplicate data from the food data recorded above through an algorithm formula. The formula for removing duplicate data is as follows: In the formula, Qccf represents the food dataset with duplicate data removed, Qcqs k represents the kth food data after removing outliers, Qcqs represents the data set after removing outliers, Qcqs m represents the mth food data after removing outliers, P i , P j Represents the feature vectors of the i-th and j-th food samples. This formula means that samples that do not have the same feature vectors as other samples are screened out from the data set after removing outliers to form a new data set.

5. The information processing method based on food detection according to claim 4 is characterized in that: The S2 removes missing values, outliers and duplicate data from the food data recorded above through an algorithm formula, and converts data of different dimensions into a unified standard. The conversion formula is as follows: In the formula, Gy represents the normalized eigenvalue, Ys represents the original eigenvalue, and Ys min 、Ys max Represents the minimum and maximum values ​​of the feature. This formula indicates that the feature value is scaled to the range of [0, 1] so that data of different dimensions can be compared under the same standard.

6. The information processing method based on food detection according to claim 5, characterized in that: The S3 extracts the color histogram from the processed food data, and the extraction formula is as follows: In the formula, H(sp) represents the histogram of color sp, I(x, y) represents the color value of image I at position (x, y), δ(sp, I(x, y)) represents the Kronecker delta function, which takes the value of 1 when I(x, y) = sp, otherwise it takes the value of 0. This formula means traversing each pixel in the image and counting the number of times each color appears.

7. The information processing method based on food detection according to claim 6 is characterized in that: S3 extracts texture features from the processed food data, and the extraction formula is as follows: In the formula, Cons represents the texture characteristics of the food image, P(i, j) represents the co-occurrence frequency of gray level i and gray level j, and contrast measures the degree of change of gray levels in the image. The larger the value, the more complex the texture.

8. The information processing method based on food detection according to claim 7, characterized in that: The S5 evaluates the performance of the model and calculates the model accuracy Mxzq. The calculation formula is as follows: In the formula, Mxzq represents the model accuracy, TP represents the number of food samples that are actually positive and correctly classified as positive, TN represents the number of food samples that are actually negative and correctly classified as negative, FP represents the number of food samples that are actually negative but mistakenly classified as positive, and FN represents the number of food samples that are actually positive but mistakenly classified as negative.

9. The information processing method based on food detection according to claim 8, characterized in that: The S5 evaluates the performance of the model and calculates the model recall rate Mxzh. The calculation formula is as follows: In the formula, Mxzh represents the model recall rate, TP represents the number of food samples that are actually positive and correctly classified as positive, and FN represents the number of food samples that are actually positive but misclassified as negative.

10. The information processing method based on food detection according to claim 9, characterized in that: The S5 evaluates the performance of the model and calculates the model F1 score Mxfs. The calculation formula is as follows: In the formula, Mxfs represents the model F1 score, Mxzq represents the model accuracy, and Mxzh represents the model recall rate.

Images