Corn ear quality detection method based on comprehensive data analysis
By combining comprehensive data analysis with image, physical and chemical features, and utilizing support vector machines and superior-inferior distance models to dynamically update the classification model, the problem of low efficiency and poor accuracy in existing corn ear quality detection technologies has been solved, achieving efficient and accurate detection results.
Patent Information
- Application Number
- CN202511559049.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-10-29
- Publication Date
- 2026-01-23
AI Technical Summary
Existing machine vision-based methods for detecting the quality of corn ears require prior knowledge, which limits the model's generalization ability and makes it unable to adapt to new samples or environments. Furthermore, traditional methods are inefficient, costly, and lack accuracy and consistency.
A comprehensive data analysis method is adopted, combining image features, physical parameters and chemical composition, and using support vector machines and superior-inferior distance models for classification. The model is optimized through a dynamic update mechanism to achieve efficient and accurate detection of corn ear quality.
In the absence of prior knowledge, it achieves efficient and accurate detection of corn ear quality, improves the accuracy and reliability of classification results, and adapts to different environments and samples.
Smart Images
Figure CN121388809A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of agricultural product quality detection, and particularly relates to a corn ear quality detection method based on comprehensive data analysis. BACKGROUND
[0002] Corn is one of the most important food crops in the world and plays a crucial role in agricultural production. The quality of corn ears is directly related to the yield, processing quality, and market value of corn. Therefore, accurately and efficiently detecting the quality of corn ears is of great significance to agricultural production and food processing industries.
[0003] Traditional methods for detecting the quality of corn ears mainly rely on human experience, which involves observing the appearance characteristics of the ears, measuring physical parameters, and performing simple chemical analysis to determine the quality of the ears. However, these methods have many limitations. First, manual detection is highly subjective, and different testers may have different standards, making it difficult to ensure the accuracy and consistency of the results. Second, manual detection is inefficient and cannot meet the needs of large-scale production. In addition, traditional methods often require destructive processing of the ears, which not only increases the cost of detection but also may result in waste of samples.
[0004] With the continuous progress of technology, machine vision and machine learning techniques are increasingly widely used in the agricultural field. Machine vision technology can quickly and accurately obtain the appearance characteristics of the ears, such as shape, color, and texture, through image processing and analysis. It can model and classify complex feature data to achieve automatic detection and evaluation of corn ear quality. However, in practical applications, existing machine vision-based corn ear quality detection methods usually require a large number of pre-defined or labeled sample data to train machine learning models. The acquisition of these sample data often requires professional knowledge and experience, such as manually labeling the quality level of samples by experts or classifying samples according to existing standards. This dependence on prior knowledge not only increases the preparation work for detection, but also may limit the generalization ability of the model, making it difficult to adapt to new samples or environments. Therefore, a corn ear quality detection method based on comprehensive data analysis is proposed. SUMMARY
[0005] The main purpose of the present application is to provide a corn ear quality detection method based on comprehensive data analysis, aiming to solve the technical problem that the existing detection based on machine vision requires prior knowledge. The method of the present application can consider various characteristic parameters such as image features, physical parameters and chemical components without prior knowledge, realize efficient and accurate detection of corn ear quality through a dynamically updated classification model, and gradually optimize the model through a dynamic updating mechanism to improve the accuracy and reliability of the classification result. The problems in the background art can be effectively solved.
[0006] To achieve the above-mentioned purpose, the technical scheme adopted by the present application is, A corn ear quality detection method based on comprehensive data analysis, comprising the following steps: Step 1: pre-selecting positive samples and negative samples from a plurality of corn ear samples to be detected, wherein the positive samples are high-quality samples, and the negative samples are low-quality samples; Step 2: acquiring characteristic parameters of the positive samples and the negative samples, respectively; Step 3: using a first classification model and a second classification model to classify the quality of a plurality of corn ear samples according to the acquired characteristic parameters, respectively, to obtain preliminary classification results; Step 4: selecting a final classification result from the preliminary classification results of the first classification model and the second classification model according to a preset selection condition; Step 5: in the classification process, if the characteristic parameters of the classified sample are better than the pre-selected positive sample, the positive sample is updated; if the characteristic parameters of the classified sample are worse than the pre-selected negative sample, the negative sample is updated; Step 6: retraining the first classification model and the second classification model according to the updated positive sample and negative sample; Step 7: repeating steps 3 to 6 until all corn ear samples to be detected are classified.
[0007] A corn ear quality detection system based on comprehensive data analysis, comprising: A sample pre-selection module for pre-selecting positive samples and negative samples from a plurality of corn ear samples to be detected; A characteristic parameter acquisition module for acquiring characteristic parameters of the positive samples and the negative samples, respectively; A first classification module for using a support vector machine model to classify the quality of a plurality of corn ear samples according to the acquired characteristic parameters, to obtain preliminary classification results; A second classification module for using a superior-inferior solution distance method model to classify the quality of a plurality of corn ear samples according to the acquired characteristic parameters, to obtain preliminary classification results; a classification result selection module configured to select a final classification result from the preliminary classification results of the support vector machine model and the outranking distance method model according to a set selection condition; a sample updating module configured to update a positive sample if a characteristic parameter of a classified sample is better than the characteristic parameter of the pre-selected positive sample, or to update a negative sample if the characteristic parameter of the classified sample is worse than the characteristic parameter of the pre-selected negative sample during the classification process; a model training module configured to retrain the support vector machine model and the outranking distance method model according to the updated positive sample and the negative sample; a control module configured to control the operation of the sample pre-selection module, the characteristic parameter acquisition module, the first classification module, the second classification module, the classification result selection module, the sample updating module, and the model training module until all the corn ear samples to be detected are classified.
[0008] A computer readable storage medium having a computer program stored thereon, wherein the computer program is executed by a processor to implement the corn ear quality detection method based on comprehensive data analysis.
[0009] Further, the characteristic parameters at least include image features and physical parameters, and further include chemical components, wherein: the image features include one or more combinations of texture, color, and shape; the physical parameters include one or more combinations of ear length, ear thickness, ear row number, row grain number, hundred-grain weight, and ear grain weight; the chemical components include one or more combinations of protein content, starch content, and fat content.
[0010] Further, in step 3, the first classification model is an SVM model, and the specific steps of using the first classification model to classify the quality of the corn ear samples include: Step 3.1: using the characteristic parameters of the positive sample and the negative sample as training data sets to train the SVM model; Step 3.2: using the trained SVM model to classify the corn ear samples to obtain preliminary classification results; Step 3.3: calculating the classification confidence of each sample according to the classification results.
[0011] Further, in step 4, the second classification model is an outranking distance method model, and the specific steps of using the second classification model to classify the quality of the corn ear samples include: Step 4.1: constructing ideal solutions and negative ideal solutions according to the characteristic parameters of the positive sample and the negative sample; Step 4.2: Calculate the distance of each sample from the ideal solution and the negative ideal solution; Step 4.3: According to the calculated distance, calculate the relative closeness of each sample; Step 4.4: According to the relative closeness, classify a plurality of corn ear samples to obtain a preliminary classification result.
[0012] Further, in step 5, the set selection condition includes: If the classification confidence of the first classification model is higher than the set threshold, the preliminary classification result of the first classification model is selected as the final classification result; If the classification confidence of the first classification model is lower than the set threshold, the preliminary classification result of the second classification model is selected as the final classification result.
[0013] Further, in step 6, the specific steps of updating the positive sample and the negative sample include: If the feature parameters of the classified sample are better than the pre-selected positive sample, the sample is used as a new positive sample; If the feature parameters of the classified sample are worse than the pre-selected negative sample, the sample is used as a new negative sample.
[0014] Further, in step 7, the specific steps of retraining the first classification model and the second classification model include: According to the updated positive sample and the negative sample, a training data set is re-constructed; Using the new training data set, the first classification model and the second classification model are retrained.
[0015] Further, in step 8, the condition for repeating steps 3 to 7 is that there are still unclassified corn ear samples.
[0016] The present application has the following beneficial effects, Compared with the prior art, the detection method proposed in the present application can effectively detect and classify the quality of corn ears through data-driven method even without prior knowledge. Compared with the prior art, the present application comprehensively considers various feature parameters such as image features, physical parameters and chemical components, and realizes efficient and accurate detection of corn ear quality through dynamically updated classification model. Compared with the prior art, the present application gradually optimizes the classification model through the dynamic updating mechanism, and further improves the accuracy and reliability of the classification result. BRIEF DESCRIPTION OF DRAWINGS
[0017] Figure 1A flowchart of a corn ear quality detection method based on comprehensive data analysis according to the present application is shown in FIG. Figure 2 A classification logic diagram of the corn ear quality detection method based on comprehensive data analysis according to the present application is shown in FIG. Figure 3 A first classification model updating flowchart in the technical solution of the present application is shown in FIG. Figure 4 A second classification model updating flowchart in the technical solution of the present application is shown in FIG. Figure 5 A structure diagram of a corn ear quality detection system based on comprehensive data analysis according to the present application is shown in FIG. DETAILED DESCRIPTION
[0018] In order to make the purposes, technical solutions and advantages of the present application clearer, the present application is further described in detail below in combination with the drawings and examples. It should be understood that the specific examples described herein are only used to explain the present application and do not limit the present application.
[0019] Example 1
[0020] Referring to Figure 1 A flowchart of a corn ear quality detection method based on comprehensive data analysis according to the present application is shown in FIG. Figure 2 A classification logic diagram of the corn ear quality detection method based on comprehensive data analysis according to the present application is shown in FIG. The method comprises the following steps: Step 1: pre-selecting positive samples and negative samples from a plurality of corn ear samples to be detected.
[0021] The positive samples are high-quality samples, and the negative samples are low-quality samples. It should be further explained that the number of positive samples and negative samples can be one or more. When multiple positive samples and negative samples are selected, the feature parameters of the multiple positive samples and negative samples are averaged, and the obtained average feature parameters are used as a reference standard for subsequent sample classification.
[0022] Step 2: obtaining feature parameters of the positive samples and the negative samples, respectively.
[0023] The feature parameters at least include image features and physical parameters, and further include chemical components, wherein: The image features include one or more combinations of texture, color and shape; The physical parameters include one or more combinations of ear length, ear thickness, ear row number, row grain number, hundred-grain weight and ear grain weight; The chemical components include one or more combinations of protein content, starch content and fat content.
[0024] In the detection of corn ear quality, the quantification of each feature parameter is a key step to achieve accurate classification and model training. Here are the specific quantification methods for each feature parameter: 1)Image Features Image features mainly include texture, color, and shape. These features can be extracted and quantified through image processing techniques.
[0025] 1.1)Texture Features Extracted through the Gray Level Co-occurrence Matrix (GLCM) method. GLCM is a statistical method that describes the spatial dependence of pixel gray values in an image. Common texture features include: Contrast: measures the clarity of texture in the image.
[0026] Homogeneity: measures the uniformity of texture in the image.
[0027] Energy: measures the intensity of texture in the image.
[0028] Correlation: measures the correlation between pixels in the image.
[0029] Quantification Method: Calculate GLCM using OpenCV or other image processing libraries.
[0030] Extract the above texture feature values.
[0031] 1.2)Color Features Color features can be extracted through methods such as color histograms or color moments. Color histograms describe the distribution of different colors in the image, while color moments provide statistical information about color distribution.
[0032] Quantification Method: Convert the image from RGB to other color spaces (such as HSV or LAB).
[0033] Calculate color histograms or color moments.
[0034] 1.3)Shape Features Shape features can be quantified through contour extraction and geometric shape analysis. Common shape features include: Area: the area of the ear.
[0035] Perimeter: the perimeter of the ear.
[0036] Aspect Ratio: the ratio of the length to the width of the ear.
[0037] Circularity: measures the similarity of the ear shape to a circle.
[0038] Quantification Method: The edge detection algorithm (such as Canny algorithm) is used to extract the contour of the ear.
[0039] The shape feature values are calculated.
[0040] 2) For the physical parameters of corn ears In one implementable way, it can be obtained by physical measurement tools. However, it needs to be measured one by one manually, which is low in efficiency.
[0041] In another implementable way, each physical parameter can be estimated by a visual detection method, specifically: 2.1) Ear length and ear thickness Image acquisition: high-resolution industrial cameras are used to take pictures from the side and end of the corn ear, ensuring clear images and complete contours of the ear.
[0042] Image preprocessing: the collected images are preprocessed for grayscale, binarization, and noise removal, etc., so as to extract the contour of the ear later.
[0043] Contour extraction and measurement: the contour of the ear is extracted by an edge detection algorithm (such as Canny algorithm). For ear length measurement, find the two end point coordinates of the ear in the image, and calculate the Euclidean distance between the two points as the ear length; for ear thickness measurement, find the ear row area in the side image of the ear, and calculate the maximum horizontal width of the ear row area as the ear thickness.
[0044] 2.2) Number of ear rows Image acquisition: take pictures from the top or bottom of the corn ear, so that the ear rows appear as parallel lines in the image.
[0045] Image preprocessing: the image is preprocessed, including grayscale, binarization, and morphological operations, to enhance the contrast of the ear rows.
[0046] Linear feature extraction and counting: linear detection algorithms such as Hough transform are used to extract straight lines in the image, and each straight line represents a row of grains. Count the number of detected straight lines to get the number of ear rows.
[0047] 2.3) Number of row grains Image acquisition: take pictures from the side of the corn ear, ensuring that each row of grains in the ear row is clearly visible.
[0048] Image preprocessing: grayscale, binarization, and possibly dilation or erosion operations are performed on the image to separate the adhered grains.
[0049] Kernel Feature Extraction and Counting: Identify the contour or region of each row of kernels through template matching, region segmentation, or object detection algorithms in deep learning, count the number of kernels in each row, and then add up all the row counts to get the total kernel count.
[0050] 2.4) Hundred Kernel Weight and Ear Kernel Weight Image Acquisition: Use high-resolution weighing sensors and cameras to simultaneously collect image and weight information of corn kernels. For hundred kernel weight, 100 kernels are first stripped from the ear, then weighed and imaged; for ear kernel weight, the entire ear is directly weighed and imaged.
[0051] Image Preprocessing: Preprocess the collected images for subsequent feature extraction.
[0052] Feature Extraction and Weight Estimation: Determine the contour and size of the ear or corn kernel through image recognition technology, and establish a relationship model between features and weight based on weight information. For hundred kernel weight, the weight of 100 kernels is directly output according to the model; for ear kernel weight, the weight of the entire ear is also output according to the model.
[0053] Visual detection method has the advantages of rapid, non-destructive and automatic in estimating the physical parameters of corn ear. However, it should be noted that in order to improve the estimation accuracy, the quality of the collected images should be ensured.
[0054] 3) For chemical composition, including protein content, starch content, fat content, etc. Then need to obtain through chemical analysis method. Specifically: 3.1) Protein Content Use Kjeldahl nitrogen determination or near-infrared spectroscopy analysis (NIRS) to determine protein content.
[0055] Quantitative method: Obtain the percentage of protein content through chemical analysis or spectral analysis.
[0056] 3.2) Starch Content Use iodine colorimetric method or near-infrared spectroscopy analysis (NIRS) to determine starch content.
[0057] Quantitative method: Obtain the percentage of starch content through chemical analysis or spectral analysis.
[0058] 3.3) Fat Content Use Soxhlet extraction method or near-infrared spectroscopy analysis (NIRS) to determine fat content.
[0059] Quantitative method: Obtain the percentage of fat content through chemical analysis or spectral analysis.
[0060] Further need to be explained, because the dimension and order of magnitude of different features may be different, the need to standardize the features. In a possible implementation, the range standardization method is adopted, and each feature parameter value is normalized to the interval [0, 1]: ; Wherein, is the jth feature parameter of the ith sample to be detected; is the jth feature parameter of the positive sample; is the jth feature parameter of the negative sample; is the jth feature parameter of the positive sample; is the jth feature parameter of the negative sample; Step 3: according to the obtained feature parameters, respectively using the first classification model and the second classification model to detect and classify the quality of a plurality of corn ear samples, and obtain the preliminary classification result.
[0061] The first classification model is an SVM model, wherein the specific steps of using the first classification model to detect and classify the quality of a plurality of corn ear samples include: Step 3.1: taking the feature parameters of the positive sample and the negative sample as the training data set, training the SVM model; Step 3.2: using the trained SVM model to classify a plurality of corn ear samples, and obtaining the preliminary classification result; Step 3.3: according to the classification result, calculating the classification confidence of each sample.
[0062] Wherein, for the calculation of the classification confidence: In a possible implementation, the confidence can be calculated based on the decision function value, specifically: using the absolute value of the decision function value as the confidence index.
[0063] In another possible implementation, the confidence can be calculated based on probability estimation, specifically: using PlattScaling or Isotonic Regression to convert the decision function value into probability estimation.
[0064] In another possible implementation, the confidence can be calculated based on distance, specifically: calculating the distance of the sample to the decision boundary as the confidence index.
[0065] The second classification model is a superior-inferior distance method model, and the specific steps of using the second classification model to detect and classify the quality of a plurality of corn ear samples include: Step 4.1: constructing ideal solution and negative ideal solution according to the feature parameters of the positive sample and the negative sample; Step 4.2: calculating the distance of each sample to the ideal solution and the negative ideal solution and , the calculation method is: ; wherein, are the maximum value and the minimum value of the jth feature parameter respectively; n is the type of feature parameters.
[0066] Step 4.3: According to the calculated distance, the relative closeness of each sample is calculated , the calculation method is: ; Step 4.4: According to the relative closeness, the corn ear samples are classified to obtain the preliminary classification result, wherein the value of the relative closeness is between [0, 1], the closer to 1 indicates that the sample is closer to the ideal solution (high-quality sample), and the closer to 0 indicates that the sample is closer to the negative ideal solution (low-quality sample); In one possible implementation, a threshold θ (for example, θ = 0.5) can be set, and the relative closeness is classified according to the relationship with θ: If ≥ θ, it is classified as a high-quality sample; If < θ, it is classified as a low-quality sample.
[0067] Step 4: According to the pre-set selection condition, the final classification result is selected from the preliminary classification results of the first classification model and the second classification model.
[0068] Wherein, the set selection condition includes: If the classification confidence of the first classification model is higher than the set threshold, the preliminary classification result of the first classification model is selected as the final classification result; If the classification confidence of the first classification model is lower than the set threshold, the preliminary classification result of the second classification model is selected as the final classification result.
[0069] Step 5: In the classification process, if the feature parameters of the classified sample are better than the pre-selected positive sample, the positive sample is updated; if the feature parameters of the classified sample are worse than the pre-selected negative sample, the negative sample is updated.
[0070] The specific steps of updating the positive sample and the negative sample include: If the feature parameters of the classified sample are better than the pre-selected positive sample, the sample is used as a new positive sample; If the feature parameters of the classified sample are worse than the pre-selected negative sample, the sample is used as a new negative sample.
[0071] It needs to be further explained that the judgment of the superior-inferior relationship between the characteristic parameters of the classified sample and the pre-selected positive sample and negative sample can be considered comprehensively in terms of multiple characteristic parameters. In a possible implementation, a linear weighting method can be used to process the multiple characteristic parameters, specifically as follows: ; wherein, is the comprehensive evaluation value of the jth characteristic parameter of the jth sample, is the comprehensive evaluation value of the jth sample, is the weighting coefficient of the jth characteristic parameter. Through the above weighting processing, the comprehensive evaluation value of the sample to be detected can be obtained. By comparing it with the comprehensive evaluation value of the pre-selected positive sample and negative sample, the superior-inferior relationship between them can be determined.
[0072] Step 6: Referring to the first classification model updating flowchart shown in Figure 3 , and the second classification model updating flowchart shown in Figure 4 , the first classification model and the second classification model are retrained according to the updated positive sample and negative sample. The specific steps include: According to the updated positive sample and negative sample, the training data set is restructured; The first classification model and the second classification model are retrained using the new training data set.
[0073] Step 7: If there are still unclassified corn ear samples, repeat steps 3 to 6 until all samples to be detected are classified.
[0074] Example 2: Referring to the structure diagram of the corn ear quality detection system based on comprehensive data analysis of the present application shown in Figure 5 , it includes: A sample pre-selection module is used to pre-select positive samples and negative samples from a plurality of corn ear samples to be detected; A characteristic parameter acquisition module is used to acquire the characteristic parameters of the positive samples and the negative samples, respectively; A first classification module is used to perform quality detection classification on a plurality of corn ear samples according to the acquired characteristic parameters using a support vector machine model to obtain a preliminary classification result; A second classification module is used to perform quality detection classification on a plurality of corn ear samples according to the acquired characteristic parameters using a superior-inferior solution distance method model to obtain a preliminary classification result; A classification result selection module is used to select the final classification result from the preliminary classification results of the support vector machine model and the superior-inferior solution distance method model according to the set selection conditions; The sample updating module is configured to update the positive sample if the characteristic parameter of the classified sample is better than the pre-selected positive sample, and update the negative sample if the characteristic parameter of the classified sample is worse than the pre-selected negative sample during the classification process. The model training module is configured to re-train the support vector machine model and the model of the optimal and worst solution distance according to the updated positive sample and negative sample. The control module is configured to control the operation of the sample pre-selection module, the characteristic parameter acquisition module, the first classification module, the second classification module, the classification result selection module, the sample updating module and the model training module until all the corn ear samples to be detected are classified.
[0075] Embodiment 3 The application further provides a computer readable storage medium, which stores a computer program, wherein the computer program can realize the steps of the corn ear quality detection method based on comprehensive data analysis when executed by a processor.
[0076] The basic principles and main features of the application and the advantages of the application are shown and described above. It should be understood by those skilled in the art that the application is not limited by the above embodiments, and the above embodiments and descriptions in the specification are only to illustrate the principles of the application. Without departing from the spirit and scope of the application, various changes and improvements can be made to the application, and these changes and improvements all fall within the scope of the claimed application. The scope of protection of the application is defined by the appended claims and their equivalents.
Claims
1. A method for detecting the quality of maize ears based on comprehensive data analysis, characterized in that, Includes the following steps: Step 1: Pre-select positive and negative samples from several corn ear samples to be tested, wherein the positive samples are high-quality samples and the negative samples are low-quality samples. Step 2: Obtain the feature parameters of the positive sample and the negative sample respectively; Step 3: Based on the obtained feature parameters, use the first classification model and the second classification model to classify and classify several corn ear samples to obtain preliminary classification results; Step 4: Select the final classification result from the preliminary classification results of the first classification model and the second classification model according to the preset selection criteria; Step 5: During the classification process, if the feature parameters of the sample being classified are better than the pre-selected positive samples, then the positive samples are updated; if the feature parameters of the sample being classified are worse than the pre-selected negative samples, then the negative samples are updated. Step 6: Retrain the first classification model and the second classification model based on the updated positive and negative samples; Step 7: Repeat steps 3 to 6 until all corn ear samples to be tested have been classified.
2. The method for detecting the quality of maize ears based on comprehensive data analysis according to claim 1, characterized in that, The feature parameters include at least image features and physical parameters, and also include chemical components, wherein: The image features include one or more of the following: texture, color, and shape; The physical parameters include one or more of the following: ear length, ear diameter, number of rows per ear, number of grains per row, weight per 100 grains, and weight per ear grain; The chemical composition includes one or more of the following: protein content, starch content, and fat content.
3. The method for detecting the quality of maize ears based on comprehensive data analysis according to claim 1, characterized in that, In step 3, the first classification model is an SVM model. The specific steps for classifying and detecting the quality of several corn ear samples using the first classification model include: Step 3.1: Use the feature parameters of the positive and negative samples as the training dataset to train the SVM model; Step 3.2: Use the trained SVM model to classify several corn ear samples to obtain preliminary classification results; Step 3.3: Calculate the classification confidence score for each sample based on the classification results.
4. The method for detecting the quality of maize ears based on comprehensive data analysis according to claim 1, characterized in that, In step 4, the second classification model is the superior-inferiority distance method. The specific steps for classifying and detecting the quality of several corn ear samples using the second classification model include: Step 4.1: Construct the ideal solution and the negative ideal solution based on the feature parameters of the positive and negative samples; Step 4.2: Calculate the distance between each sample and the ideal solution and the negative ideal solution; Step 4.3: Calculate the relative proximity of each sample based on the calculated distance; Step 4.4: Classify several corn ear samples according to their relative proximity to obtain preliminary classification results.
5. The method for detecting the quality of maize ears based on comprehensive data analysis according to claim 1, characterized in that, In step 5, the selection criteria include: If the classification confidence of the first classification model is higher than the set threshold, then the preliminary classification result of the first classification model is selected as the final classification result. If the classification confidence of the first classification model is lower than a set threshold, the preliminary classification result of the second classification model is selected as the final classification result.
6. The method for detecting the quality of maize ears based on comprehensive data analysis according to claim 1, characterized in that, In step 6, the specific steps for updating the positive and negative samples include: If the feature parameters of the sample to be classified are better than those of the pre-selected positive samples, then the sample is taken as the new positive sample. If the feature parameters of the sample being classified are inferior to those of the pre-selected reverse sample, then the sample is used as the new reverse sample.
7. The method for detecting the quality of maize ears based on comprehensive data analysis according to claim 1, characterized in that, In step 7, the specific steps for retraining the first classification model and the second classification model include: Reconstruct the training dataset based on the updated positive and negative samples; The first classification model and the second classification model are retrained using a new training dataset.
8. The method for detecting the quality of maize ears based on comprehensive data analysis according to claim 1, characterized in that, In step 8, the condition for repeating steps 3 to 7 is that there are still unclassified corn ear samples.
9. A corn ear quality detection system based on comprehensive data analysis, characterized in that, include: The sample pre-selection module is used to pre-select positive and negative samples from several corn ear samples to be tested; The feature parameter acquisition module is used to acquire the feature parameters of the positive sample and the negative sample respectively; The first classification module is used to classify and detect the quality of several corn ear samples using a support vector machine model based on the acquired feature parameters, and to obtain preliminary classification results. The second classification module is used to classify several corn ear samples based on the acquired feature parameters using the superior-inferior solution distance method model, and obtain preliminary classification results. The classification result selection module is used to select the final classification result from the preliminary classification results of the support vector machine model and the superior-inferior solution distance method model according to the set selection conditions. The sample update module is used to update positive samples during the classification process if the feature parameters of the classified sample are better than the pre-selected positive samples; and to update negative samples if the feature parameters of the classified sample are worse than the pre-selected negative samples. The model training module is used to retrain the support vector machine model and the superior-inferior solution distance method model based on the updated positive and negative samples. The control module is used to control the operation of the sample pre-selection module, feature parameter acquisition module, first classification module, second classification module, classification result selection module, sample update module and model training module until all corn ear samples to be detected are classified.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the corn ear quality detection method based on comprehensive data analysis as described in any one of claims 1 to 8.