A method and device for detecting pea seed quality based on infrared spectroscopy

Abstract

The invention discloses a pea seed quality detection method and device based on infrared spectroscopy, and the invention relates to the technical field of seed screening. The method comprises the following steps: collecting infrared spectra of sample pea seeds with known quality parameters, performing denoising and enhancement preprocessing, and establishing a neural network prediction model in combination with quality parameters (water content, crude fiber, fat, starch, protein); performing the same preprocessing on the infrared spectra of the seeds to be tested, inputting the trained model, and predicting their quality parameters; calculating the nutrient richness factor and water absorption expansion potential factor based on the predicted values, and calculating the color uniformity coefficient in combination with the color channel value extracted from the surface image; calculating the quality index by comprehensively analyzing information such as color uniformity, circularity, and surface roughness, and generating a quality judgment result by comparing it with a preset judgment threshold, which can quickly output the quality judgment result of pea seeds and meet the needs of modern agriculture for high-throughput detection and intelligent analysis.

Description

Technical Field

[0001] The invention relates to the technical field of seed screening, and in particular to a pea seed quality detection method and device based on infrared spectroscopy. Background Art

[0002] With the continuous development of agricultural production, seed quality detection and evaluation technology has received widespread attention. The quality of seeds directly determines the growth, yield and quality of crops. For pea seeds, quality parameters such as water content, crude fiber content, fat content, starch content and protein content are important indicators to measure their nutritional value and processing performance. These components not only affect the germination rate and growth potential of peas, but are also directly related to their suitability as food or industrial raw materials. However, traditional seed detection methods usually rely on chemical analysis or manual evaluation, which usually have the disadvantages of being time-consuming, costly and destructive to samples. In addition, these methods often require professional operation and are difficult to meet the needs of modern agriculture for rapid and non-destructive testing.

[0003] Infrared spectroscopy has been widely used in food quality testing and agricultural product analysis due to its rapid, non-destructive and no need for complex sample preparation. Infrared spectroscopy can provide detailed data on the chemical composition of seeds by detecting the molecular vibration information of samples. On this basis, the rapid development of neural network technology, especially deep learning, in recent years has provided new technical means for seed quality detection. Neural networks can learn complex nonlinear relationships through a large amount of training data and are suitable for processing high-dimensional nonlinear features such as infrared spectral data. However, it is still insufficient to evaluate the comprehensive quality of seeds based solely on the prediction model of chemical parameters. For example, the appearance characteristics of seeds (such as color uniformity, roundness and surface roughness) are also important factors in measuring seed quality. If these appearance characteristics are ignored and seed quality is evaluated based only on chemical composition parameters, the final judgment may be one-sided and the accuracy may be reduced.

[0004] In addition, modern breeding and seed markets have placed higher demands on the comprehensive quality of pea seeds. For example, the color uniformity of pea seeds directly affects their sensory value in the market, while the water absorption and swelling potential is closely related to the germination speed and food processing characteristics of the seeds. Therefore, an evaluation method that is not only fast, non-destructive and high-throughput, but also capable of comprehensively considering multiple chemical components and physical parameters is needed to provide users with comprehensive seed quality evaluation results.

[0005] In the prior art, the publication number CN119246455A discloses a Cistanche deserticola seed screening method based on infrared spectroscopy, which includes seed classification data determination, infrared spectroscopy detection and data processing and analysis; the seed classification data determination specifically includes: purity determination, thousand-grain weight determination, seed vigor determination and moisture content determination. Through this method, it is obtained that: Cistanche deserticola seeds of different grades are significantly correlated with infrared spectral characteristic peaks, wherein the characteristic peaks have a high positive correlation with thousand-grain weight, seed vigor and moisture content, indicating that the absorbance of the corresponding band of Cistanche deserticola seed samples is an important indicator that can reflect seed quality; absorbance is significantly correlated with thousand-grain weight and seed vigor, proving the effectiveness and accuracy of infrared spectroscopy in evaluating seed quality, vigor and thousand-grain weight. However, this method only detects the seed chemical composition of seeds through infrared spectroscopy technology, relies on a single chemical composition as a criterion, and does not consider the appearance and shape characteristics, thereby reducing the accuracy and effectiveness of the screening system.

[0006] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not constitute the prior art that is already known to one of ordinary skill in the art. Summary of the invention

[0007] The object of the present invention is to provide a pea seed quality detection method and device based on infrared spectroscopy to solve the problems raised in the above background technology.

[0008] To achieve the above object, the present invention provides the following technical solutions:

[0009] A method for detecting pea seed quality based on infrared spectroscopy, the specific steps comprising:

[0010] Collecting infrared spectra of several sample pea seeds with known quality parameters, and preprocessing the collected infrared spectra of the sample pea seeds to obtain several training infrared spectra, wherein the preprocessing includes signal denoising and enhancement, and the quality parameters include water content, crude fiber content, fat content, starch content and protein content;

[0011] A neural network prediction model is established, the training infrared spectrum is used as the input of the neural network prediction model, and the quality parameter corresponding to each training infrared spectrum is used as a label to train the neural network prediction model to obtain a quality parameter prediction model;

[0012] Performing infrared detection on the pea seeds to be detected, collecting the corresponding initial infrared spectrum, preprocessing the initial infrared spectrum to obtain a target infrared spectrum, inputting the target infrared spectrum into the trained quality parameter prediction model, and the model outputs the quality parameter prediction value of the pea seeds to be detected;

[0013] Based on the obtained predicted values ​​of the quality parameters of the pea seeds to be tested, the nutritional richness factor and the water absorption and swelling potential factor are calculated, and the surface image of the pea seeds to be tested is collected. Based on the surface image of the pea seeds to be tested, the color channel value of each pixel in the seed image is extracted, and the color uniformity coefficient of the pea seeds to be tested is calculated based on the obtained color channel values;

[0014] According to the obtained color uniformity coefficient, nutritional richness factor and water absorption swelling potential factor, combined with the roundness and surface roughness of the pea seeds to be tested, the quality index of the pea seeds to be tested is comprehensively calculated, and the obtained quality index of the pea seeds to be tested is compared with the quality judgment threshold, and the corresponding pea seed quality judgment result is generated according to the obtained comparison result.

[0015] Furthermore, the method for performing noise reduction and enhancement processing on the infrared spectrum of the collected sample pea seeds is: using a wavelet transform denoising method to denoise the infrared spectrum image, and the specific steps of the wavelet transform denoising method include: decomposing the distortion-corrected image through a wavelet transform to obtain wavelet coefficients of the image at different scales and directions; performing threshold processing on the wavelet coefficients, setting the low-amplitude wavelet coefficients to zero, and retaining the high-amplitude wavelet coefficients; performing an inverse transform on the wavelet coefficients after the threshold processing, reconstructing the processed coefficients into an image, and completing the image denoising processing;

[0016] Bilateral filtering is used to enhance the details of infrared spectral images. The formula for the specific filtering transformation is:

[0017] ;

[0018] In the formula, is the coordinate vector in the image coordinate system, is the coordinate vector The gray value at Gray value The gray value after bilateral filtering transformation is are all Gaussian functions, where The formula is:

[0019] ;

[0020] ;

[0021] In the formula, is the coordinate vector in the image coordinate system, is the coordinate vector The gray value at and They are The standard deviation of .

[0022] Furthermore, a quality parameter prediction model is established based on a convolutional neural network, wherein the convolutional neural network consists of an input layer, a convolutional layer, a pooling layer, a fully connected layer and an output layer, wherein the activation function in the convolutional layer is function, The specific expression of the function is:

[0023] ;

[0024] in, Indicates The corresponding convolutional layers, It means the The corresponding convolutional layer The feature map eigenvalues;

[0025] For the fully connected layer, the number of neurons in the fully connected layer is set to 32, the initial neural network learning rate is set to 0.001, and the number of training rounds is 100;

[0026] The trained quality parameter prediction model takes as input the infrared spectrum image of pea seeds, and outputs the corresponding quality parameter prediction values, including the moisture content prediction value, crude fiber content prediction value, fat content prediction value, starch content prediction value and protein content prediction value.

[0027] Furthermore, based on the predicted values ​​of the quality parameters of the pea seeds to be tested, the nutritional richness factor and the water absorption and swelling potential factor are calculated, wherein the formula for calculating the nutritional richness factor is:

[0028] ;

[0029] In the formula, is the nutritional richness factor of the pea seeds to be tested, is the predicted value of protein content of the pea seeds to be tested, is the predicted value of starch content of the pea seeds to be tested, is the predicted value of fat content of the pea seeds to be tested, is the predicted value of crude fiber content of pea seeds to be tested;

[0030] The formula for calculating the water absorption expansion potential factor is:

[0031] ;

[0032] In the formula, is the water absorption and swelling potential factor of the pea seeds to be tested, is the predicted value of moisture content of the pea seeds to be tested.

[0033] Further, the color uniformity coefficient of the pea seeds to be detected is calculated based on the obtained color channel value, wherein the color uniformity coefficient of the pea seeds to be detected is calculated according to the formula:

[0034] ;

[0035] In the formula, is the color uniformity coefficient of the pea seeds to be tested, is the deviation of the red channel component of the pea seed surface image to be detected, is the deviation of the green channel component, is the deviation of the blue channel component;

[0036] The specific calculation formula for the deviation of the red channel, green channel and blue channel components of the pea seed surface image to be detected is:

[0037] ;

[0038] ;

[0039] ;

[0040] In the formula, is the total number of pixels of the surface image, For the The red channel value of each pixel, For the The green channel value of each pixel, For the The blue channel value of each pixel, , and They are the standard red channel value, standard green channel value, and standard blue channel value of pea seeds, is the index of the pixel point of the surface image, where .

[0041] Further, according to the obtained color uniformity coefficient, nutrient richness factor and water absorption expansion potential factor, combined with the roundness and surface roughness of the pea seeds to be tested, the quality index of the pea seeds to be tested is comprehensively calculated, wherein the formula for calculating the quality index of the pea seeds to be tested is:

[0042] ;

[0043] In the formula, is the quality index of the pea seeds to be tested, is the roundness of the pea seeds to be tested, is the surface roughness of the pea seeds to be tested, and are weight coefficients respectively, among which and and Both are greater than 0.

[0044] Furthermore, the obtained quality index of the pea seeds to be detected is compared with the quality judgment threshold, and the corresponding pea seed quality judgment result is generated according to the obtained comparison result, wherein the specific judgment logic is:

[0045] when When the pea seeds are detected, they are judged to be high-quality seeds, indicating that the pea seeds to be tested are suitable for sowing or breeding;

[0046] when When the quality of the pea seeds is medium, it is judged as medium quality seeds, which means that the pea seeds to be tested can be used for some non-critical agricultural or industrial purposes;

[0047] when When the pea seeds are detected, they are judged as low-quality seeds, indicating that the quality of the pea seeds to be tested is poor and cannot meet normal planting or processing requirements;

[0048] in It is the preset quality judgment threshold.

[0049] The present invention also provides a pea seed quality detection device based on infrared spectroscopy, wherein the pea seed quality detection device based on infrared spectroscopy is used to perform the above-mentioned pea seed quality detection method based on infrared spectroscopy, comprising:

[0050] A sample data processing module, used for collecting infrared spectra of several sample pea seeds with known quality parameters, and preprocessing the collected infrared spectra of the sample pea seeds to obtain several training infrared spectra, wherein the preprocessing includes signal denoising and enhancement processing, and the quality parameters include water content, crude fiber content, fat content, starch content and protein content;

[0051] A neural network training module is used to establish a neural network prediction model, use the training infrared spectrum as the input of the neural network prediction model, and use the quality parameter corresponding to each training infrared spectrum as a label to train the neural network prediction model to obtain a quality parameter prediction model;

[0052] The quality parameter prediction module is used to perform infrared detection on the pea seeds to be detected, collect the corresponding initial infrared spectrum, pre-process the initial infrared spectrum to obtain the target infrared spectrum, input the target infrared spectrum into the trained quality parameter prediction model, and the model outputs the quality parameter prediction value of the pea seeds to be detected;

[0053] A characteristic parameter characterization module is used to calculate the nutritional richness factor and the water absorption expansion potential factor based on the obtained quality parameter prediction value of the pea seeds to be tested, collect the surface image of the pea seeds to be tested, extract the color channel value of each pixel in the seed image based on the surface image of the pea seeds to be tested, and calculate the color uniformity coefficient of the pea seeds to be tested based on the obtained color channel value;

[0054] The comprehensive quality judgment module is used to comprehensively calculate the quality index of the pea seeds to be tested based on the obtained color uniformity coefficient, nutritional richness factor and water absorption expansion potential factor, combined with the roundness and surface roughness of the pea seeds to be tested, compare the obtained quality index of the pea seeds to be tested with the quality judgment threshold, and generate the corresponding pea seed quality judgment result according to the obtained comparison result.

[0055] Compared with the prior art, the present invention has the following beneficial effects:

[0056] Firstly, the infrared spectra of pea seeds were collected to extract the chemical composition characteristics in the samples, and the spectral data were optimized by signal denoising and enhancement processing to improve the accuracy and stability of the data. By adopting the neural network prediction model, a nonlinear mapping relationship between infrared spectra and seed quality parameters can be established to achieve rapid prediction of seed moisture content, crude fiber, fat, starch and protein content. Compared with traditional linear regression or chemometric methods, it has higher flexibility and accuracy and can better adapt to the complexity of infrared spectral data. Secondly, a comprehensive evaluation was further conducted in combination with the physical properties and appearance parameters of the seeds. By analyzing the surface images of pea seeds, the color channel values ​​were extracted and the color uniformity coefficient was calculated, which can reflect the visual consistency and sensory quality of the seeds. At the same time, the water absorption and swelling potential factor was calculated, which is closely related to the germination ability and subsequent processing performance of the seeds. This makes up for the shortcomings of the traditional detection methods that ignore the appearance and physical characteristics. Finally, a more comprehensive quality evaluation was achieved by introducing the comprehensive quality index and comprehensive calculation. This method is not only fast and non-destructive, but also can analyze chemical composition and physical properties at the same time. It can quickly output the quality judgment results of pea seeds and meet the needs of modern agriculture for high-throughput detection and intelligent analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 It is a schematic diagram of the overall method flow of the present invention;

[0058] Figure 2 It is a schematic diagram of the overall structure of the device of the present invention. DETAILED DESCRIPTION

[0059] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific embodiments.

[0060] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The words "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0061] Example:

[0062] See also Figure 1 , the present invention provides a technical solution:

[0063] A method for detecting pea seed quality based on infrared spectroscopy, the specific steps comprising:

[0064] Step 1: Collect infrared spectra of several sample pea seeds with known quality parameters, and preprocess the collected infrared spectra of the sample pea seeds to obtain several training infrared spectra, wherein the preprocessing includes signal denoising and enhancement processing, and the quality parameters include water content, crude fiber content, fat content, starch content and protein content.

[0065] Moisture content is an important indicator of seeds, which will directly affect the storage and germination performance of seeds. The moisture content is mainly reflected by the absorption peak of the OH bond. The infrared spectrum is very sensitive to moisture content in the 1400-1900nm region (the first water absorption band) and the 2500-2600nm region (the second water absorption band). The moisture content of seeds can be quantitatively analyzed through the absorption intensity of these characteristic bands.

[0066] Protein is an important nutrient in pea seeds and is mainly composed of amino acids. The characteristic absorption bands of protein are concentrated at 1500-1700nm and 2100-2300nm (absorption of NH and CH bonds). The protein content can be predicted by the intensity and shape of these characteristic bands.

[0067] Starch is the main carbohydrate in pea seeds, which is mainly composed of glucose polymerization. The infrared spectrum of starch absorbs characteristic bands from CH bonds and OH bonds, concentrated in 2100-2300nm and 2500-2600nm. The starch content of seeds can be estimated by quantitatively analyzing the absorption intensity of these bands.

[0068] The fat content in pea seeds is low, but the fat content can be detected by the absorption of CH bonds. The characteristic bands are mainly located at 1700-1800nm ​​and 2300-2500nm. The fat content can be quantitatively analyzed through the changes in these absorption peaks.

[0069] Crude fiber is mainly composed of cellulose, hemicellulose and pectin, and is the structural component of pea seeds. The absorption band of crude fiber is concentrated in the range of 1900-2400nm (the characteristic band of OH and CH bonds). Spectral data can be used to reflect the relative content of crude fiber.

[0070] The method for performing noise reduction and enhancement processing on the infrared spectrum of the collected sample pea seeds is: using the wavelet transform denoising method to denoise the infrared spectrum image, and the specific steps of the wavelet transform denoising method include: decomposing the distortion-corrected image through wavelet transform to obtain the wavelet coefficients of the image at different scales and directions; performing threshold processing on the wavelet coefficients, setting the low-amplitude wavelet coefficients to zero, and retaining the high-amplitude wavelet coefficients; performing inverse transformation on the wavelet coefficients after threshold processing, reconstructing the processed coefficients into an image, and completing the image denoising processing;

[0071] Bilateral filtering is used to enhance the details of infrared spectral images. The formula for the specific filtering transformation is:

[0072] ;

[0073] In the formula, is the coordinate vector in the image coordinate system, is the coordinate vector The gray value at Gray value The gray value after bilateral filtering transformation is are all Gaussian functions, where The formula is:

[0074] ;

[0075] ;

[0076] In the formula, is the coordinate vector in the image coordinate system, is the coordinate vector The gray value at and They are The standard deviation of .

[0077] Step 2: Establish a neural network prediction model, use the training infrared spectrum as the input of the neural network prediction model, and use the quality parameters corresponding to each training infrared spectrum as a label to train the neural network prediction model to obtain a quality parameter prediction model.

[0078] A quality parameter prediction model is established based on a convolutional neural network, wherein the convolutional neural network consists of an input layer, a convolutional layer, a pooling layer, a fully connected layer and an output layer, wherein the activation function in the convolutional layer is function, The specific expression of the function is:

[0079] ;

[0080] in, Indicates The corresponding convolutional layers, It means the The corresponding convolutional layer The feature map eigenvalues;

[0081] For the fully connected layer, the number of neurons in the fully connected layer is set to 32, the initial neural network learning rate is set to 0.001, and the number of training rounds is 100;

[0082] The trained quality parameter prediction model takes as input the infrared spectrum image of pea seeds, and outputs the corresponding quality parameter prediction values, including the moisture content prediction value, crude fiber content prediction value, fat content prediction value, starch content prediction value and protein content prediction value.

[0083] The biggest advantage of convolutional neural networks is their powerful automated feature extraction capabilities. In infrared spectral data, different spectral bands have different sensitivities to specific quality parameters (such as water content, protein content, etc.). Traditional machine learning methods usually require manual selection or design of features (such as principal component analysis, chemometric methods, etc.), which may lead to insufficient feature extraction or loss of important information. However, CNN can automatically learn local features in infrared spectral data through convolution operations without relying on manual intervention, thereby mining deeper and more fine-grained feature information and significantly improving prediction accuracy.

[0084] Infrared spectral data is usually high-dimensional and covers a large number of bands. When processing high-dimensional data, the traditional fully connected neural network (FCNN) will cause a sharp increase in parameters, high model complexity, and easy overfitting. CNN can effectively reduce the number of parameters through the convolution kernel (i.e., weight sharing mechanism) while capturing local patterns in spectral data.

[0085] Step 3: Perform infrared detection on the pea seeds to be tested, collect the corresponding initial infrared spectrum, preprocess the initial infrared spectrum to obtain the target infrared spectrum, input the target infrared spectrum into the trained quality parameter prediction model, and the model outputs the predicted value of the quality parameter of the pea seeds to be tested.

[0086] The preprocessing method is the same as above and will not be described in detail here.

[0087] Step 4: Based on the predicted values ​​of the quality parameters of the pea seeds to be tested, the nutritional richness factor and the water absorption and swelling potential factor are calculated, and the surface image of the pea seeds to be tested is collected. Based on the surface image of the pea seeds to be tested, the color channel value of each pixel in the seed image is extracted, and the color uniformity coefficient of the pea seeds to be tested is calculated based on the obtained color channel values.

[0088] Based on the predicted values ​​of the quality parameters of the pea seeds to be tested, the nutritional richness factor and the water absorption and swelling potential factor are calculated, wherein the formula for calculating the nutritional richness factor is:

[0089] ;

[0090] In the formula, is the nutritional richness factor of the pea seeds to be tested, is the predicted value of protein content of the pea seeds to be tested, is the predicted value of starch content of the pea seeds to be tested, is the predicted value of fat content of the pea seeds to be tested, is the predicted value of crude fiber content of pea seeds to be tested;

[0091] It should be noted that the nutritional richness factor of the pea seeds to be tested is By comprehensively analyzing the starch content and protein content of the pea seeds to be tested, the nutrition contained in the pea seeds is characterized. The larger the value, the richer the nutrition contained in the pea seeds, and the more suitable it is for sowing, development and processing, that is, the higher the variety of the pea seeds.

[0092] Protein is one of the main nutrients in pea seeds and an important indicator of its nutritional value. Protein is an essential nutrient for the human body and is involved in the growth, repair and immune function of the body. Therefore, the more protein content, the higher the nutritional value. Therefore, the predicted value of protein content in the pea seeds to be tested is The nutritional richness factor of pea seeds to be tested Proportional, using exponential function The contribution of protein content to the overall nutritional factor is amplified in a nonlinear form. This means that as the protein content increases, its effect on nutritional richness increases exponentially rather than linearly, reflecting the prominent position of protein content in the nutritional evaluation of pea seeds, and is suitable for emphasizing the superiority of pea varieties with higher protein content in nutritional value.

[0093] Starch is the main carbohydrate source of pea seeds, which is the core component for providing energy. The higher the starch content, the greater the energy supply potential of pea seeds. Fat is another important nutrient in peas. Although the content is low, it has a high calorie density and can provide a long-lasting energy source for the body. Fat is also involved in some important metabolic functions, such as the absorption of fat-soluble vitamins. Therefore, the predicted value of starch content in the pea seeds to be tested is and prediction of fat content of pea seeds to be tested The nutritional richness factor of the pea seeds to be tested Proportional, using Grouping starch and fat together reflects their synergistic contribution to overall nutritional value. The square operation can amplify the impact of starch content on the overall nutritional value, making it more weighted, and adding the fat content prediction value Finally, the contributions of the two are balanced by the square root operation. The square root can reduce the unbalanced impact of an extremely high value of a component on the result while retaining the combined contribution of the two.

[0094] Crude fiber is an important component of pea seeds, but it mainly plays a role in filling and promoting digestion, and does not directly provide energy or nutrition to the body. Excessive crude fiber content may reduce the digestion and absorption rate of food, especially the absorption efficiency of starch, protein and fat. Therefore, the crude fiber content of the pea seeds to be tested is predicted. is set as the denominator to represent its limiting effect on nutrient richness.

[0095] Overall through the logarithmic function The calculation results will present a smoother growth trend, making the contributions of different nutrients more balanced.

[0096] The formula for calculating the water absorption expansion potential factor is:

[0097] ;

[0098] In the formula, is the water absorption and swelling potential factor of the pea seeds to be tested, is the predicted value of moisture content of the pea seeds to be tested.

[0099] It should be noted that the water absorption and swelling properties of pea seeds are reflected by integrating the three key parameters of pea seeds: water content, starch content and crude fiber content. Used to evaluate the storage stability and processing adaptability of pea seeds (such as the ability to swell during soaking, germination or milling). The higher the value, the more suitable the seeds are for professional sowing or deep processing.

[0100] Predicted moisture content of pea seeds to be tested Represents the initial moisture state of the seeds. The higher the moisture content, the easier it is for the seeds to absorb more water and swell. Seeds with high initial moisture content have stronger swelling potential. It is placed directly in the molecular position of the formula to reflect its positive effect on the water absorption and expansion potential.

[0101] Starch is the main water-absorbing component of seeds. Starch granules have strong water absorption and swelling capacity and can absorb a large amount of water and swell, thus Play a significant positive role in improving the denominator by setting Indicates a proportional relationship.

[0102] Crude fiber has a certain limiting effect on the water absorption and swelling capacity of seeds. Crude fiber is a structural component whose main function is to provide physical support rather than to absorb water and swell. The presence of crude fiber may limit the space for seeds to absorb water and swell or hinder water from entering the seeds. Therefore, the predicted value of crude fiber content in the pea seeds to be tested is and Inversely proportional, in order to reflect the inhibitory effect of crude fiber, a logarithmic relationship is introduced into the formula This reflects that the restrictive effect of crude fiber on water absorption expansion does not increase linearly: a lower crude fiber content may have a significant inhibitory effect on the expansion performance, but as the crude fiber content increases, its restrictive effect gradually becomes gentle.

[0103] The color uniformity coefficient of the pea seeds to be detected is calculated based on the obtained color channel value, wherein the color uniformity coefficient of the pea seeds to be detected is calculated based on the formula:

[0104] ;

[0105] In the formula, is the color uniformity coefficient of the pea seeds to be tested, is the deviation of the red channel component of the pea seed surface image to be detected, is the deviation of the green channel component, is the deviation of the blue channel component;

[0106] The specific calculation formula for the deviation of the red channel, green channel and blue channel components of the pea seed surface image to be detected is:

[0107] ;

[0108] ;

[0109] ;

[0110] In the formula, is the total number of pixels of the surface image, For the The red channel value of each pixel, For the The green channel value of each pixel, For the The blue channel value of each pixel, , and They are the standard red channel value, standard green channel value, and standard blue channel value of pea seeds, is the index of the pixel point of the surface image, where .

[0111] It should be noted that The larger the value, the more uniform the color and the smaller the difference from the standard color channel. The standard red channel value, standard green channel value, and standard blue channel value of pea seeds , and For details, please refer to relevant data and make settings based on expert experience.

[0112] Step 5: Based on the obtained color uniformity coefficient, nutritional richness factor and water absorption swelling potential factor, combined with the roundness and surface roughness of the pea seeds to be tested, the quality index of the pea seeds to be tested is comprehensively calculated, and the obtained quality index of the pea seeds to be tested is compared with the quality judgment threshold value. According to the obtained comparison result, the corresponding pea seed quality judgment result is generated.

[0113] According to the obtained color uniformity coefficient, nutrient richness factor and water absorption expansion potential factor, combined with the roundness and surface roughness of the pea seeds to be tested, the quality index of the pea seeds to be tested is comprehensively calculated, wherein the formula for calculating the quality index of the pea seeds to be tested is:

[0114] ;

[0115] In the formula, is the quality index of the pea seeds to be tested, is the roundness of the pea seeds to be tested, is the surface roughness of the pea seeds to be tested, and are weight coefficients respectively, among which and and Both are greater than 0.

[0116] The quality index of pea seeds to be tested It indicates the comprehensive quality of planting. The higher the value, the higher the quality of the pea seeds to be tested.

[0117] Nutrient Richness Factor Reflects the nutritional value of pea seeds and is the core indicator for describing the intrinsic quality of seeds. For high-quality seeds, nutritional value is dominant and is the indicator that consumers or growers care about most. The square weighting is used for the nutritional richness factor. , indicating that the contribution of this factor to pea seed quality is nonlinear and has an amplification effect. It reflects the physical properties of pea seeds and is an important indicator for evaluating the edible quality of pea seeds. It also reflects the adaptability of peas to growth or processing. Therefore, it is related to the quality index of the pea seeds to be tested. Proportional, through Comprehensive indication of the quality of nutritional growth.

[0118] Roundness Indicates the regularity of the seed shape. Seeds that are close to round are generally considered to be high-quality seeds. The higher the roundness, the better the visual quality, so it is related to the quality index of the pea seeds to be tested. The formula is calculated by the sum of color uniformity and circularity. Using the Logarithmic Function , the contribution of appearance indicators is characterized nonlinearly, avoiding the excessive influence of extreme values ​​of a certain factor on the quality index, making the results more stable and scientific.

[0119] The higher the surface roughness of the pea seeds to be tested, the more likely it is that there are defects such as cracks and shrinkage, which may lead to a decrease in the quality of the pea seeds. Inversely proportional.

[0120] Since the effect of appearance index on quality index is weaker than that of nutrition type parameter, and and All of them are greater than 0, which reasonably reflects that the importance of appearance indicators is not as good as nutritional and functional indicators, but they are still important references in quality evaluation.

[0121] The circularity can be calculated by an empirical formula, and the surface roughness can be obtained by scanning with a laser diffractometer. Both are conventional technical means and will not be described in detail here.

[0122] The obtained quality index of the pea seeds to be tested is compared with the quality judgment threshold, and the corresponding pea seed quality judgment result is generated according to the obtained comparison result, wherein the specific judgment logic is:

[0123] when When the pea seeds are detected, they are judged to be high-quality seeds, indicating that the pea seeds to be tested are suitable for sowing or breeding;

[0124] when When the pea seeds to be tested are judged to be medium-quality seeds, it means that the pea seeds to be tested can be used for some non-critical agricultural or industrial purposes; the non-critical agriculture includes leisure agriculture, pet and animal breeding, such as farmhouses, agricultural sightseeing parks, picking gardens, etc., whose main purpose is to provide leisure and entertainment services and has no direct relationship with food production.

[0125] when When the pea seeds are detected, they are judged as low-quality seeds, indicating that the quality of the pea seeds to be tested is poor and cannot meet normal planting or processing requirements;

[0126] in It is the preset quality judgment threshold.

[0127] See also Figure 2 The present invention also provides a pea seed quality detection device based on infrared spectroscopy, wherein the pea seed quality detection device based on infrared spectroscopy is used to perform the above-mentioned pea seed quality detection method based on infrared spectroscopy, comprising:

[0128] A sample data processing module, used for collecting infrared spectra of several sample pea seeds with known quality parameters, and preprocessing the collected infrared spectra of the sample pea seeds to obtain several training infrared spectra, wherein the preprocessing includes signal denoising and enhancement processing, and the quality parameters include water content, crude fiber content, fat content, starch content and protein content;

[0129] A neural network training module is used to establish a neural network prediction model, use the training infrared spectrum as the input of the neural network prediction model, and use the quality parameter corresponding to each training infrared spectrum as a label to train the neural network prediction model to obtain a quality parameter prediction model;

[0130] The quality parameter prediction module is used to perform infrared detection on the pea seeds to be detected, collect the corresponding initial infrared spectrum, pre-process the initial infrared spectrum to obtain the target infrared spectrum, input the target infrared spectrum into the trained quality parameter prediction model, and the model outputs the quality parameter prediction value of the pea seeds to be detected;

[0131] A characteristic parameter characterization module is used to calculate the nutritional richness factor and the water absorption expansion potential factor based on the obtained quality parameter prediction value of the pea seeds to be tested, collect the surface image of the pea seeds to be tested, extract the color channel value of each pixel in the seed image based on the surface image of the pea seeds to be tested, and calculate the color uniformity coefficient of the pea seeds to be tested based on the obtained color channel value;

[0132] The comprehensive quality judgment module is used to comprehensively calculate the quality index of the pea seeds to be tested based on the obtained color uniformity coefficient, nutritional richness factor and water absorption expansion potential factor, combined with the roundness and surface roughness of the pea seeds to be tested, compare the obtained quality index of the pea seeds to be tested with the quality judgment threshold, and generate the corresponding pea seed quality judgment result according to the obtained comparison result.

[0133] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0134] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented by software, the above embodiments may be implemented in whole or in part in the form of a computer program product. Those skilled in the art may appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein may be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed by hardware or software methods depends on the specific application and design constraints of the technical solution.

[0135] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, and may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0136] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. A method for detecting pea seed quality based on infrared spectroscopy, characterized in that: The specific steps include: Collecting infrared spectra of several sample pea seeds with known quality parameters, and preprocessing the collected infrared spectra of the sample pea seeds to obtain several training infrared spectra, wherein the preprocessing includes signal denoising and enhancement, and the quality parameters include water content, crude fiber content, fat content, starch content and protein content; A neural network prediction model is established, the training infrared spectrum is used as the input of the neural network prediction model, and the quality parameter corresponding to each training infrared spectrum is used as a label to train the neural network prediction model to obtain a quality parameter prediction model; Performing infrared detection on the pea seeds to be detected, collecting the corresponding initial infrared spectrum, preprocessing the initial infrared spectrum to obtain a target infrared spectrum, inputting the target infrared spectrum into the trained quality parameter prediction model, and the model outputs the quality parameter prediction value of the pea seeds to be detected; Based on the obtained predicted values ​​of the quality parameters of the pea seeds to be tested, the nutritional richness factor and the water absorption and swelling potential factor are calculated, and the surface image of the pea seeds to be tested is collected. Based on the surface image of the pea seeds to be tested, the color channel value of each pixel in the seed image is extracted, and the color uniformity coefficient of the pea seeds to be tested is calculated based on the obtained color channel values; According to the obtained color uniformity coefficient, nutritional richness factor and water absorption swelling potential factor, combined with the roundness and surface roughness of the pea seeds to be tested, the quality index of the pea seeds to be tested is comprehensively calculated, and the obtained quality index of the pea seeds to be tested is compared with the quality judgment threshold, and the corresponding pea seed quality judgment result is generated according to the obtained comparison result.

2. The method for detecting pea seed quality based on infrared spectroscopy according to claim 1, characterized in that: The method for performing noise reduction and enhancement processing on the infrared spectrum of the collected sample pea seeds is: using the wavelet transform denoising method to denoise the infrared spectrum image, and the specific steps of the wavelet transform denoising method include: decomposing the distortion-corrected image through wavelet transform to obtain the wavelet coefficients of the image at different scales and directions; performing threshold processing on the wavelet coefficients, setting the low-amplitude wavelet coefficients to zero, and retaining the high-amplitude wavelet coefficients; performing inverse transformation on the wavelet coefficients after threshold processing, reconstructing the processed coefficients into an image, and completing the image denoising processing; Bilateral filtering is used to enhance the details of infrared spectral images. The formula for the specific filtering transformation is: ； In the formula, is the coordinate vector in the image coordinate system, is the coordinate vector The gray value at Gray value The gray value after bilateral filtering transformation is are all Gaussian functions, where The formula is: ； ； In the formula, is the coordinate vector in the image coordinate system, is the coordinate vector The gray value at and They are The standard deviation of .

3. The method for detecting pea seed quality based on infrared spectroscopy according to claim 2, characterized in that: A quality parameter prediction model is established based on a convolutional neural network, wherein the convolutional neural network consists of an input layer, a convolutional layer, a pooling layer, a fully connected layer and an output layer, wherein the activation function in the convolutional layer is function, The specific expression of the function is: ； in, Indicates The corresponding convolutional layers, It means the The corresponding convolutional layer The feature map eigenvalues; For the fully connected layer, the number of neurons in the fully connected layer is set to 32, the initial neural network learning rate is set to 0.001, and the number of training rounds is 100; The trained quality parameter prediction model takes as input the infrared spectrum image of pea seeds, and outputs the corresponding quality parameter prediction values, including the moisture content prediction value, crude fiber content prediction value, fat content prediction value, starch content prediction value and protein content prediction value.

4. The method for detecting pea seed quality based on infrared spectroscopy according to claim 1, characterized in that: Based on the predicted values ​​of the quality parameters of the pea seeds to be tested, the nutritional richness factor and the water absorption and swelling potential factor are calculated, wherein the formula for calculating the nutritional richness factor is: ； In the formula, is the nutritional richness factor of the pea seeds to be tested, is the predicted value of protein content of the pea seeds to be tested, is the predicted value of starch content of the pea seeds to be tested, is the predicted value of fat content of the pea seeds to be tested, is the predicted value of crude fiber content of pea seeds to be tested; The formula for calculating the water absorption expansion potential factor is: ； In the formula, is the water absorption and swelling potential factor of the pea seeds to be tested, is the predicted value of moisture content of the pea seeds to be tested.

5. The method for detecting pea seed quality based on infrared spectroscopy according to claim 4, characterized in that: The color uniformity coefficient of the pea seeds to be detected is calculated based on the obtained color channel value, wherein the color uniformity coefficient of the pea seeds to be detected is calculated based on the formula: ； In the formula, is the color uniformity coefficient of the pea seeds to be tested, is the deviation of the red channel component of the pea seed surface image to be detected, is the deviation of the green channel component, is the deviation of the blue channel component; The specific calculation formula for the deviation of the red channel, green channel and blue channel components of the pea seed surface image to be detected is: ； ； ； In the formula, is the total number of pixels of the surface image, For the The red channel value of each pixel, For the The green channel value of each pixel, For the The blue channel value of each pixel, , and They are the standard red channel value, standard green channel value, and standard blue channel value of pea seeds, is the index of the pixel point of the surface image, where .

6. The method for detecting pea seed quality based on infrared spectroscopy according to claim 5, characterized in that: According to the obtained color uniformity coefficient, nutrient richness factor and water absorption expansion potential factor, combined with the roundness and surface roughness of the pea seeds to be tested, the quality index of the pea seeds to be tested is comprehensively calculated, wherein the formula for calculating the quality index of the pea seeds to be tested is: ； In the formula, is the quality index of the pea seeds to be tested, is the roundness of the pea seeds to be tested, is the surface roughness of the pea seeds to be tested, and are weight coefficients respectively, among which and and Both are greater than 0.

7. The method for detecting pea seed quality based on infrared spectroscopy according to claim 6, characterized in that: The obtained quality index of the pea seeds to be tested is compared with the quality judgment threshold, and the corresponding pea seed quality judgment result is generated according to the obtained comparison result, wherein the specific judgment logic is: when When the pea seeds are detected, they are judged to be high-quality seeds, indicating that the pea seeds to be tested are suitable for sowing or breeding; when When the quality of the pea seeds is determined to be medium, the pea seeds to be tested can be used for non-critical agricultural or industrial purposes; when When the pea seeds are detected, they are judged as low-quality seeds, indicating that the quality of the pea seeds to be tested is poor and cannot meet normal planting or processing requirements; in It is the preset quality judgment threshold.

8. A pea seed quality detection device based on infrared spectroscopy, characterized in that: The pea seed quality detection device based on infrared spectroscopy is used to execute the pea seed quality detection method based on infrared spectroscopy according to any one of claims 1 to 7, comprising: A sample data processing module, used for collecting infrared spectra of several sample pea seeds with known quality parameters, and preprocessing the collected infrared spectra of the sample pea seeds to obtain several training infrared spectra, wherein the preprocessing includes signal denoising and enhancement processing, and the quality parameters include water content, crude fiber content, fat content, starch content and protein content; A neural network training module is used to establish a neural network prediction model, use the training infrared spectrum as the input of the neural network prediction model, and use the quality parameter corresponding to each training infrared spectrum as a label to train the neural network prediction model to obtain a quality parameter prediction model; The quality parameter prediction module is used to perform infrared detection on the pea seeds to be detected, collect the corresponding initial infrared spectrum, pre-process the initial infrared spectrum to obtain the target infrared spectrum, input the target infrared spectrum into the trained quality parameter prediction model, and the model outputs the quality parameter prediction value of the pea seeds to be detected; A characteristic parameter characterization module is used to calculate the nutritional richness factor and the water absorption expansion potential factor based on the obtained quality parameter prediction value of the pea seeds to be tested, collect the surface image of the pea seeds to be tested, extract the color channel value of each pixel in the seed image based on the surface image of the pea seeds to be tested, and calculate the color uniformity coefficient of the pea seeds to be tested based on the obtained color channel value; The comprehensive quality judgment module is used to comprehensively calculate the quality index of the pea seeds to be tested based on the obtained color uniformity coefficient, nutritional richness factor and water absorption expansion potential factor, combined with the roundness and surface roughness of the pea seeds to be tested, compare the obtained quality index of the pea seeds to be tested with the quality judgment threshold, and generate the corresponding pea seed quality judgment result according to the obtained comparison result.

Images