A detection method and device for antioxidant activity of oat seeds based on cellulase

Abstract

The present invention relates to the technical field of deep learning, and discloses a method for detecting the antioxidant activity of oat seeds based on cellulase, including: crushing a plurality of oat seeds of different types into an oat flour set, performing cellulose hydrolysis on each oat flour in the oat flour set by using cellulase to obtain an oat hydrolyzate set, performing spectral imaging on the oat hydrolyzate set to obtain an oat spectral atlas set; performing keto-phenol detection on each oat hydrolyzate in the oat hydrolyzate set to obtain an oat component parameter group set, performing antioxidant detection on the oat hydrolyzate set by using the free radical scavenging method to obtain an antioxidant data set; training an antioxidant activity model by using the oat spectral atlas set and the oat component parameter group set to obtain an antioxidant analysis model; and analyzing the antioxidant activity of the to-be-detected oat seeds by using the antioxidant analysis model. The present invention also provides a device for detecting the antioxidant activity of oat seeds based on cellulase. The present invention can improve the efficiency of analyzing the antioxidant activity of oat seeds.

Description

Technical Field

[0001] The present invention relates to the technical field of deep learning, and particularly relates to a method and device for detecting the antioxidant activity of oat seeds based on cellulase. Background Art

[0002] With the development of the economy and the improvement of living standards, more and more people's health awareness has increased accordingly. In order to improve physical ability, people have started to take the consumption of foods with antioxidant activity as a trend. Oat seeds are considered an important source of natural antioxidants. In order to improve the economic value of oat seeds, it is necessary to detect and analyze the antioxidant activity of oat seeds.

[0003] The existing detection techniques for the antioxidant activity of oat seeds are mostly based on biochemical antioxidant activity detection methods. By detecting the oxidation efficiency between oat seed samples and various chemical substances, the antioxidant capacity of oat seeds is determined. In practical applications, when detecting the antioxidant activity of multiple experimental samples, the biochemical antioxidant activity detection method requires a large number of repetitive chemical reactions and the determination of various complex experimental data. The repetitive and complex experimental data determination may lead to low efficiency in detecting the antioxidant activity of oat seeds. Summary of the Invention

[0004] The present invention provides a method and device for detecting the antioxidant activity of oat seeds based on cellulase, and its main purpose is to solve the problem of low efficiency in detecting the antioxidant activity of oat seeds.

[0005] To achieve the above object, a method for detecting the antioxidant activity of oat seeds based on cellulase provided by the present invention includes:

[0006] Crushing multiple oat seeds of different types into an oat flour set, hydrolyzing the cellulose of each oat flour in the oat flour set by using cellulase to obtain an oat hydrolyzate set, and performing spectral imaging on the oat hydrolyzate set to obtain an oat spectral atlas;

[0007] Performing glyco-ketone-phenol detection on each oat hydrolyzate in the oat hydrolyzate set to obtain an oat component parameter group set, and performing antioxidant detection on the oat hydrolyzate set by using the free radical scavenging method to obtain an antioxidant data set;

[0008] Selecting each oat spectral image in the oat spectral atlas as a target oat spectral image one by one, selecting the oat component parameter group corresponding to the target oat spectral image in the oat component parameter group set as a target component parameter group, and sequentially performing image denoising, wavelength calibration, and image enhancement operations on the target oat spectral image to obtain a target standard spectral image;

[0009] Vectorize the target ingredient parameter group into target ingredient features, downsample the target standard spectral image into target spectral features according to the target ingredient features, fuse the target ingredient features and the target spectral features into target oat features, and use all the target oat features and the antioxidant data set to train a preset antioxidant activity model to obtain an antioxidant analysis model. Among them, the step of fusing the target ingredient features and the target spectral features into target oat features includes: performing a global pooling operation on the target ingredient features to obtain dimension-reduced ingredient features; performing a global pooling operation on the target spectral features to obtain dimension-reduced spectral features; using the following multi-dimensional oat feature fusion algorithm to perform feature fusion on the dimension-reduced ingredient features and the dimension-reduced spectral features to obtain target oat features:

[0010]

[0011] where Z refers to the target oat feature, i refers to the feature dimension serial number, E refers to the total dimension of the feature vectors of the dimension-reduced ingredient features, and the total dimension of the feature vectors of the dimension-reduced ingredient features is the same as that of the dimension-reduced spectral features. softmax is a normalization function, Q i is the feature vector of the i-th dimension feature in the dimension-reduced ingredient features, α, β, and γ are the preset fusion coefficient matrices of the multi-dimensional oat feature fusion algorithm, w() is a dimension function, refers to Q i the dimension of the β vector, refers to R i the dimension of the β vector, and T is the transpose symbol;

[0012] Hydrolyze the oats to be tested into hydrolyzates to be tested, generate the spectral images to be tested and the parameter groups of the ingredients to be tested of the hydrolyzates to be tested respectively, generate the oat features to be tested according to the spectral images to be tested and the parameter groups of the ingredients to be tested, and use the antioxidant analysis model to analyze the antioxidant data of the oat features to be tested, and use the analyzed antioxidant data as the antioxidant activity of the oats to be tested.

[0013] Optionally, the step of crushing multiple oats of different varieties into an oat flour set includes:

[0014] Collect multiple oats of different varieties into an oat variety set, and select one oat variety in the oat variety set as the target oat variety one by one;

[0015] Perform cleaning and drying operations on the target oat variety in sequence to obtain a target clean oat variety;

[0016] Perform crushing and sieving operations on the target clean oat variety in sequence to obtain a target primary oat flour;

[0017] Screen out oat flour of a preset quality from the target primary oat flour and collect all the target oat flour into an oat flour set.

[0018] Optionally, using cellulase to perform cellulose hydrolysis on each oat flour in the oat flour set to obtain an oat hydrolyzate set, including:

[0019] Select the oat flour in the oat flour set one by one as the target oat flour, stir and dissolve the target oat flour to obtain an oat flour solution;

[0020] Filter out a standard oat liquid from the oat flour solution, and use cellulase to perform a fiber hydrolysis reaction on the standard oat liquid to obtain a hydrolysis reaction liquid;

[0021] Filter and precipitate a primary hydrolyzate from the hydrolysis reaction liquid, perform a drying operation on the primary hydrolyzate to obtain an oat hydrolyzate, and collect all the oat hydrolyzates into an oat hydrolyzate set.

[0022] Optionally, perform a glyco-ketone-phenol detection on each oat hydrolyzate in the oat hydrolyzate set to obtain an oat component parameter group set, including:

[0023] Select the oat hydrolyzate in the oat hydrolyzate set one by one as the target oat hydrolyzate, and use the precipitation acid method to detect the total polysaccharide content in the target oat hydrolyzate;

[0024] Use the aluminum salt colorimetric method to detect the total flavonoid content in the target oat hydrolyzate;

[0025] Use the phenol-sulfuric acid method to detect the reducing sugar content in the target oat hydrolyzate;

[0026] Use the sodium salt colorimetric method to detect the total phenol content in the target oat hydrolyzate;

[0027] Collect the total polysaccharide content, the total flavonoid content, the reducing sugar content, and the total phenol content into an oat component parameter group of the target oat hydrolyzate, and collect all the oat component parameter groups into an oat component parameter group set.

[0028] Optionally, perform an antioxidant detection on the oat hydrolyzate set using the free radical scavenging method to obtain an antioxidant data set, including:

[0029] Select the oat hydrolyzate in the oat hydrolyzate set one by one as the target oat hydrolyzate, and dissolve the target oat hydrolyzate into a target oat solvent group with different concentrations;

[0030] Use the pre-prepared free radical solution to react with each target oat solvent in the target oat solvent group respectively to obtain a target oat reaction solvent group;

[0031] Use a spectrophotometer to detect the corresponding target absorbance group of the target oat reaction solvent group respectively;

[0032] Calculate the free radical scavenging rate of the target oat hydrolyzate according to the target absorbance group and the preset positive control group;

[0033] Take the free radical scavenging rate as the antioxidant data of the target oat hydrolyzate, and collect all the antioxidant data into an antioxidant data set.

[0034] Optionally, the operations of image denoising, wavelength calibration, and image enhancement are sequentially performed on the target oat spectral image to obtain a target standard spectral image, including:

[0035] Use the decomposition and reconstruction method to perform threshold denoising on the target oat spectral image to obtain a target denoised spectral image;

[0036] Perform baseline correction on the target denoised spectral image to obtain a target baseline spectral image;

[0037] Perform wavelength calibration on the target baseline spectral image according to the pre-acquired sample spectral image to obtain a target calibrated spectral image;

[0038] Generate a grayscale histogram of the target calibrated spectral image, and perform grayscale equalization operation on the target calibrated spectral image according to the grayscale histogram to obtain a target standard spectral image.

[0039] Optionally, the use of the decomposition and reconstruction method to perform threshold denoising on the target oat spectral image to obtain a target denoised spectral image includes:

[0040] Use the following multi-layer decomposition algorithm to perform filtering decomposition on the target oat spectral image to obtain a spectral decomposition hierarchy set and a decomposition coefficient set corresponding to the spectral decomposition hierarchy set:

[0041]

[0042] where x j,k refers to the k-th decomposition coefficient of the j-th spectral decomposition hierarchy in the spectral decomposition hierarchy set, s j refers to the scale coefficient of the j-th spectral decomposition hierarchy, n refers to the signal scale serial number of the target oat spectral image, N refers to the signal scale size of the target oat spectral image, h n refers to the first filtering coefficient of the target oat spectral image at the n-th signal scale, d j refers to the detail coefficient of the j-th spectral decomposition hierarchy, g nrefers to the second filtering coefficient of the target oat spectral image at the nth signal scale;

[0043] Select each spectral decomposition level in the spectral decomposition level set as the target spectral decomposition level one by one, take the decomposition coefficient corresponding to the target spectral decomposition level in the decomposition coefficient set as the target decomposition coefficient, and extract the hierarchical noise feature and hierarchical signal feature from the target spectral decomposition level in turn;

[0044] Fuse the hierarchical noise feature and the hierarchical signal feature into a hierarchical feature, and generate a hierarchical threshold according to the hierarchical feature;

[0045] Judge whether the hierarchical threshold is less than the target decomposition coefficient;

[0046] If not, return to the step of selecting each spectral decomposition level in the spectral decomposition level set as the target spectral decomposition level one by one;

[0047] If so, add the target spectral decomposition level to the preset reconstruction level set. When the target spectral decomposition level is the last level in the spectral decomposition level set, obtain the standard reconstruction level set;

[0048] Reconstruct the standard reconstruction level set by using the following multi-layer reconstruction algorithm to obtain the target denoised spectral image:

[0049]

[0050] where, x j+1,k refers to the kth decomposition coefficient of the (j + 1)th spectral decomposition level in the spectral decomposition level set, s j refers to the scale coefficient of the jth spectral decomposition level, n refers to the signal scale serial number of the target oat spectral image, N refers to the signal scale size of the target oat spectral image, h k-2n refers to the first filtering coefficient of the target oat spectral image at the (k - 2n)th signal scale, d j refers to the detail coefficient of the jth spectral decomposition level, g k-2n refers to the second filtering coefficient of the target oat spectral image at the (k - 2n)th signal scale.

[0051] Optionally, the downsampling the target standard spectral image into the target spectral feature according to the target component feature includes:

[0052] Extract the total dimension of the feature vector from the target component feature;

[0053] Use a multi-level convolutional layer to extract various spectral spatial features from the target standard spectral image;

[0054] Perform frequency-domain transformation on the target standard spectral image to obtain the target spectral frequency domain, and extract various spectral frequency domain features from the target spectral frequency domain;

[0055] Integrate all spectral spatial features and all spectral frequency domain features into a primary spectral feature set;

[0056] Perform clustering operation on the primary spectral feature set with the total dimension of the feature vector as the number of clustering clusters to obtain a spectral feature class set;

[0057] Perform downsampling operation on each spectral feature class in the spectral feature class set to obtain multiple weighted spectral center features, and integrate all the weighted spectral center features into the target spectral features.

[0058] Optionally, training the preset antioxidant activity model with all the target oat features and the antioxidant data set to obtain an antioxidant analysis model includes:

[0059] Analyze the primary antioxidant data corresponding to the target oat features one by one using the preset antioxidant activity model, and integrate all the primary antioxidant data into a primary antioxidant data set;

[0060] Calculate the loss value of the antioxidant activity model according to the primary antioxidant data set and the antioxidant data set;

[0061] Determine whether the loss value is greater than a preset loss value threshold;

[0062] If so, update the model parameters of the antioxidant activity model according to the loss value, and return to the step of analyzing the primary antioxidant data corresponding to the target oat features one by one using the preset antioxidant activity model;

[0063] If not, use the updated antioxidant activity model as the antioxidant analysis model.

[0064] To solve the above problems, the present invention also provides an antioxidant activity detection device for oat seeds based on cellulase, and the device includes:

[0065] A spectral imaging module, configured to crush multiple oat seeds of different types into an oat flour set, perform cellulose hydrolysis on each oat flour in the oat flour set using cellulase to obtain an oat hydrolyzate set, and perform spectral imaging on the oat hydrolyzate set to obtain an oat spectral atlas;

[0066] A parameter detection module, configured to perform keto-phenol detection on each oat hydrolyzate in the oat hydrolyzate set to obtain an oat component parameter group set, and perform antioxidant detection on the oat hydrolyzate set using the free radical scavenging method to obtain an antioxidant data set;

[0067] An image enhancement module, configured to select oat spectral pictures in the oat spectral atlas one by one as target oat spectral pictures, select the oat component parameter group corresponding to the target oat spectral picture in the oat component parameter group set as the target component parameter group, and perform image denoising, wavelength calibration, and image enhancement operations on the target oat spectral pictures in sequence to obtain target standard spectral pictures;

[0068] A model training module, configured to vectorize the target component parameter group into target component features, downsample the target standard spectral pictures into target spectral features according to the target component features, fuse the target component features and the target spectral features into target oat features, and use all the target oat features and the antioxidant data set to train a preset antioxidant activity model to obtain an antioxidant analysis model, where fusing the target component features and the target spectral features into target oat features includes: performing a global pooling operation on the target component features to obtain dimension-reduced component features; performing a global pooling operation on the target spectral features to obtain dimension-reduced spectral features; using the following multi-dimensional oat feature fusion algorithm to perform feature fusion on the dimension-reduced component features and the dimension-reduced spectral features to obtain target oat features:

[0069]

[0070] where Z refers to the target oat feature, i refers to the feature dimension serial number, E refers to the total dimension of the feature vectors of the dimension-reduced component features, and the total dimension of the feature vectors of the dimension-reduced component features is the same as the total dimension of the feature vectors of the dimension-reduced spectral features, softmax is a normalization function, Q i is the feature vector of the i-th dimension feature in the dimension-reduced component features, α, β, γ are fusion coefficient matrices preset by the multi-dimensional oat feature fusion algorithm, w() is a dimension function, refers to Q i the dimension of the β vector, refers to R i the dimension of the β vector, T is the transpose symbol;

[0071] An activity detection module, configured to hydrolyze the to-be-detected oat seeds into to-be-detected hydrolyzates, generate to-be-detected spectral pictures and to-be-detected component parameter groups of the to-be-detected hydrolyzates respectively, generate to-be-detected oat features according to the to-be-detected spectral pictures and the to-be-detected component parameter groups, analyze the antioxidant data of the to-be-detected oat features by using the antioxidant analysis model, and use the analyzed antioxidant data as the antioxidant activity of the to-be-detected oat seeds.

[0072] The embodiment of the present invention grinds a plurality of oat seeds of different types into an oat flour set, and uses cellulase to hydrolyze the cellulose of each oat flour in the oat flour set to obtain an oat hydrolyzate set. The cellulase can be used to dissolve the cellulose of the oat seeds, thereby improving the antioxidant activity value of the oat, making the experimental data clearer and easier to observe. By performing spectral imaging on the oat hydrolyzate set to obtain an oat spectrum atlas, material imaging of each oat hydrolyzate in the oat hydrolyzate set can be obtained, thereby obtaining more accurate measurement data of the antioxidant activity. By performing sugar, ketone and phenol detection on each oat hydrolyzate in the oat hydrolyzate set, the antioxidant activity of the oat hydrolyzate can be obtained. By using the oat component parameter set, parameter information of antioxidant-related substances such as sugars, ketones, phenols, etc. can be obtained. By using the free radical scavenging method to perform antioxidant detection on the oat hydrolyzate set, an antioxidant data set can be obtained. The real measured antioxidant data set corresponding to the oat hydrolyzate set can be analyzed according to a unified scale, thereby facilitating subsequent model training. By performing image denoising, wavelength calibration and image enhancement operations on the target oat spectral image in sequence, a target standard spectral image can be obtained, which can make the target oat spectral image clearer and the details more obvious, while retaining more spectral details, thereby enhancing the coverage of subsequent target spectral features.

[0073] By downsampling the target standard spectrum image into a target spectrum feature according to the target component feature, it is possible to ensure that the feature dimensions of the target component feature and the target spectrum feature are the same, facilitating subsequent feature fusion according to the structure. By fusing the target component feature and the target spectrum feature into a target oat feature, the feature dimension can be reduced, the magnitude of the calculation can be reduced, and the efficiency of the calculation can be improved. By using all the target oat features and the antioxidant data set to train a preset antioxidant activity model, an antioxidant analysis model is obtained, which can improve the accuracy of the antioxidant activity analysis. By using the antioxidant analysis model to analyze the analytical antioxidant data of the oat feature to be tested, the analytical antioxidant data is used as the antioxidant activity of the oat species to be tested. The antioxidant data can be analyzed based on the spectral characteristics of the cellulase-enzymed oat species to be tested and the characteristics of substances such as saccharide phenols in combination with a deep learning algorithm, thereby improving the efficiency of antioxidant data analysis. Therefore, the cellulase-based oat seed antioxidant activity detection method and device proposed in the present invention can solve the problem of low efficiency when performing oat seed antioxidant activity detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] Figure 1 A schematic diagram of a process for detecting antioxidant activity of oat seeds based on cellulase according to one embodiment of the present invention;

[0075] Figure 2 A schematic diagram of a process for detecting sugar ketone phenols according to an embodiment of the present invention;

[0076] Figure 3 Schematic flowchart of generating target spectral features provided by an embodiment of the present invention;

[0077] Figure 4 Functional module diagram of an antioxidant activity detection device for oat seeds based on cellulase provided by an embodiment of the present invention;

[0078] The realization, functional features and advantages of the present invention will be further described with reference to the embodiments and the accompanying drawings. Specific embodiments

[0079] It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0080] An embodiment of the present application provides an antioxidant activity detection method for oat seeds based on cellulase. The execution subject of the antioxidant activity detection method for oat seeds based on cellulase includes but is not limited to at least one of electronic devices such as a server, a terminal, etc. that can be configured to execute the method provided by the embodiment of the present application. In other words, the antioxidant activity detection method for oat seeds based on cellulase can be executed by software or hardware installed on a terminal device or a server device, and the software can be a blockchain platform. The server includes but is not limited to: a single server, a server cluster, a cloud server, or a cloud server cluster, etc. The server can be an independent server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, Content Delivery Network (CDN), and big data and artificial intelligence platforms.

[0081] Refer to Figure 1 As shown, it is a schematic flowchart of an antioxidant activity detection method for oat seeds based on cellulase provided by an embodiment of the present invention. In this embodiment, the antioxidant activity detection method for oat seeds based on cellulase includes:

[0082] S1. Crush multiple oat seeds of different types into an oat flour set, perform cellulose hydrolysis on each oat flour in the oat flour set using cellulase to obtain an oat hydrolyzate set, and perform spectral imaging on the oat hydrolyzate set to obtain an oat spectral atlas.

[0083] In the embodiment of the present invention, the multiple oat seeds of different types refer to multiple oat seed samples from different regions and different types, and the oat seeds can be oats of varieties such as short oats, desert oats, sand oats, small-grained naked oats, incomplete oats, long-haired oats, long-glumed oats, overweight oats, Canadian oats, Damascus oats, and creeping oats.

[0084] In an embodiment of the present invention, the step of crushing multiple oat seeds of different types into an oat flour set includes:

[0085] Pool multiple oat seeds of different types into an oat seed set, and select one oat seed from the oat seed set one by one as the target oat seed;

[0086] Perform cleaning and drying operations on the target oat seed in sequence to obtain a target clean oat seed;

[0087] Perform crushing and sieving operations on the target clean oat seed in sequence to obtain a target primary oat flour;

[0088] Screen out oat flour of a preset quality from the target primary oat flour as the target oat flour, and pool all the target oat flours into an oat flour set.

[0089] Specifically, the target oat seed can be cleaned with pure water, and the cleaned target oat seed can be dried by methods such as reduced-pressure drying or spray drying to obtain a target clean oat seed.

[0090] In an embodiment of the present invention, the target clean oat seed can be crushed by a mortar grinding method, an ultrasonic crushing method or a ball mill grinding method, and the target clean oat seed can be sieved by a screening method or a comb tooth method to obtain a target primary oat flour.

[0091] Specifically, the step of performing cellulose hydrolysis on each oat flour in the oat flour set by using cellulase to obtain an oat hydrolyzate set includes:

[0092] Select the oat flour in the oat flour set one by one as the target oat flour, and stir and dissolve the target oat flour to obtain an oat flour solution;

[0093] Filter out a standard oat liquid from the oat flour solution, and perform a fiber hydrolysis reaction on the standard oat liquid by using cellulase to obtain a hydrolysis reaction liquid;

[0094] Filter and precipitate a primary hydrolyzate from the hydrolysis reaction liquid, perform a drying operation on the primary hydrolyzate to obtain an oat hydrolyzate, and pool all the oat hydrolyzates into an oat hydrolyzate set.

[0095] Specifically, the target oat flour can be stirred and dissolved with distilled water on a constant temperature shaker to obtain an oat flour solution. The step of performing a fiber hydrolysis reaction on the standard oat liquid by using cellulase to obtain a hydrolysis reaction liquid means that in an environment with a temperature of 50-60 °C and a pH value of 4.5-5.5, a quantitative cellulase is added to a quantitative standard oat liquid and reacted for 24 hours to obtain a hydrolysis reaction liquid.

[0096] Specifically, a filter paper or a filter can be used to filter the oat flour solution and the hydrolysis reaction solution, and anhydrous ethanol can be used to precipitate the primary hydrolyzate.

[0097] Specifically, methods such as an Atomic Force Microscope (AFM) and a Scanning Electron Microscope (SEM) can be used to perform spectral imaging on the oat hydrolyzate set to obtain an oat spectral atlas.

[0098] In the embodiment of the present invention, by crushing multiple oat seeds of different types into an oat flour set, using cellulase to perform cellulose hydrolysis on each oat flour in the oat flour set to obtain an oat hydrolyzate set, the cellulase can be used to dissolve the cellulose of the oat seeds, thereby increasing the antioxidant activity value of oats, making the experimental data clearer and more convenient to observe. By performing spectral imaging on the oat hydrolyzate set to obtain an oat spectral atlas, the material imaging of each oat hydrolyzate in the oat hydrolyzate set can be obtained, thereby obtaining more accurate measurement data of the antioxidant activity.

[0099] S2. Perform keto-phenol detection on each oat hydrolyzate in the oat hydrolyzate set to obtain an oat component parameter group set, and perform antioxidant detection on the oat hydrolyzate set using the free radical scavenging method to obtain an antioxidant data set.

[0100] In the embodiment of the present invention, the oat component parameter group set is a set composed of multiple oat component parameter groups, and the oat component parameter group contains parameters such as the total polysaccharide content, total flavonoid content, reducing sugar content, and total phenol content corresponding to one oat hydrolyzate.

[0101] In the embodiment of the present invention, with reference to Figure 2 As shown, the performing keto-phenol detection on each oat hydrolyzate in the oat hydrolyzate set to obtain an oat component parameter group set includes:

[0102] S21. Select each oat hydrolyzate in the oat hydrolyzate set as the target oat hydrolyzate one by one, and use the precipitation acid method to detect the total polysaccharide content from the target oat hydrolyzate;

[0103] S22. Use the aluminum salt colorimetric method to detect the total flavonoid content from the target oat hydrolyzate;

[0104] S23. Use the phenol-sulfuric acid method to detect the reducing sugar content from the target oat hydrolyzate;

[0105] S24. Use the sodium salt colorimetric method to detect the total phenol content from the target oat hydrolyzate;

[0106] S25. Aggregate the total polysaccharide content, the total flavonoid content, the reducing sugar content, and the total phenol content into an oat component parameter group of the target oat hydrolysate, and aggregate all the oat component parameter groups into an oat component parameter group set.

[0107] Specifically, the precipitation acid method can be the total polysaccharide content determination method of national standard GB / T22416 - 2008. Detecting the total flavonoid content from the target oat hydrolysate by the aluminum salt colorimetric method means that after co - precipitating the target oat hydrolysate and ethanol, using 2% potassium aluminate solution as the reactant, measuring the absorbance value with an ultraviolet spectrophotometer after a series of reactions, and calculating the total flavonoid content by comparing with the standard curve.

[0108] Specifically, detecting the reducing sugar content from the target oat hydrolysate by the phenol - sulfuric acid method means mixing the target oat hydrolysate and phenol - sulfuric acid, adding sulfuric acid after a fixed time, measuring the absorbance value with an ultraviolet spectrophotometer after the reaction, and calculating the reducing sugar content by comparing with the standard curve.

[0109] Specifically, detecting the total phenol content from the target oat hydrolysate by the sodium salt colorimetric method can be the Folin–Ciocalteu method, that is, mixing the target oat hydrolysate with Folin–Ciocalteu reagent and sodium carbonate solution, measuring the absorbance value with an ultraviolet spectrophotometer after the reaction, and calculating the total phenol content by comparing with the standard curve.

[0110] Specifically, performing antioxidant detection on the oat hydrolysate set by the free radical scavenging method to obtain an antioxidant data set, including:

[0111] Select the oat hydrolysates in the oat hydrolysate set one by one as the target oat hydrolysates, and dissolve the target oat hydrolysates into a target oat solvent group with different concentrations;

[0112] React the pre - prepared free radical solution with each target oat solvent in the target oat solvent group respectively to obtain a target oat reaction solvent group;

[0113] Use a spectrophotometer to detect the corresponding target absorbance group of the target oat reaction solvent group respectively;

[0114] Calculate the free radical scavenging rate of the target oat hydrolysate according to the target absorbance group and the preset positive control group;

[0115] Take the free radical scavenging rate as the antioxidant data of the target oat hydrolysate, and aggregate all the antioxidant data into an antioxidant data set.

[0116] Specifically, the target oat hydrolysate can be dissolved in a methanol solution to form target oat solvent groups with different concentrations; the free radical solution can be a DPPH free radical solution or an ABTS free radical.

[0117] Specifically, the positive control group refers to the absorbance group after reacting with a free radical solution of a positive sample solvent such as vitamin C or hawthorn powder with the same concentration as the target oat solvent group. The free radical scavenging rate of the target oat hydrolysate calculated based on the target absorbance group and the preset positive control group means that the difference between the ratio of each absorbance in the target absorbance group to the blank absorbance and 1 is used as the initial scavenging rate, and the initial scavenging rate is scaled by a magnification factor according to the ratio relationship between the target absorbance group and the preset positive control group to obtain the free radical scavenging rate.

[0118] In the embodiments of the present invention, by performing keto-phenol detection on each oat hydrolysate in the oat hydrolysate set, an oat component parameter set can be obtained, and parameter information of substances such as carbohydrates, ketones, and phenols related to antioxidant properties can be acquired. By using the free radical scavenging method to perform antioxidant detection on the oat hydrolysate set, an antioxidant data set can be obtained, and the true measured antioxidant data set corresponding to the oat hydrolysate set can be analyzed on a unified scale, thereby facilitating subsequent model training.

[0119] S3. Select each oat spectral image in the oat spectral atlas as a target oat spectral image one by one, select the oat component parameter set corresponding to the target oat spectral image in the oat component parameter set as the target component parameter set, and perform image denoising, wavelength calibration, and image enhancement operations on the target oat spectral image in sequence to obtain a target standard spectral image.

[0120] In the embodiments of the present invention, the selection of the oat component parameter set corresponding to the target oat spectral image in the oat component parameter set as the target component parameter set means that the oat hydrolysate corresponding to the target oat spectral image is used as the target oat hydrolysate, and the oat component parameter set corresponding to the target oat hydrolysate in the oat component parameter set is used as the target component parameter set.

[0121] In the embodiments of the present invention, the sequential performance of image denoising, wavelength calibration, and image enhancement operations on the target oat spectral image to obtain a target standard spectral image includes:

[0122] Performing threshold denoising on the target oat spectral image using the decomposition and reconstruction method to obtain a target denoised spectral image;

[0123] Performing baseline correction on the target denoised spectral image to obtain a target baseline spectral image;

[0124] Perform wavelength calibration on the target baseline spectral image according to the pre-acquired sample spectral image to obtain a target calibrated spectral image;

[0125] Generate a grayscale histogram of the target calibrated spectral image, and perform grayscale equalization operation on the target calibrated spectral image according to the grayscale histogram to obtain a target standard spectral image.

[0126] Specifically, the polynomial fitting method can be used to perform baseline correction on the target denoised spectral image to obtain a target baseline spectral image, and the sample spectral image and the target baseline spectral image can be aligned in spectral wavelength by the cross-correlation alignment method to obtain a target calibrated spectral image.

[0127] In the embodiment of the present invention, the use of the decomposition and reconstruction method to perform threshold denoising on the target oat spectral image to obtain a target denoised spectral image includes:

[0128] Perform filtering decomposition on the target oat spectral image by using the following multi-layer decomposition algorithm to obtain a spectral decomposition level set and a decomposition coefficient set corresponding to the spectral decomposition level set:

[0129]

[0130] where x j,k refers to the k-th decomposition coefficient of the j-th spectral decomposition level in the spectral decomposition level set, s j refers to the scale coefficient of the j-th spectral decomposition level, n refers to the signal scale serial number of the target oat spectral image, N refers to the signal scale size of the target oat spectral image, h n refers to the first filtering coefficient of the target oat spectral image at the n-th signal scale, d j refers to the detail coefficient of the j-th spectral decomposition level, g n refers to the second filtering coefficient of the target oat spectral image at the n-th signal scale;

[0131] Select each spectral decomposition level in the spectral decomposition level set as the target spectral decomposition level, use the decomposition coefficient corresponding to the target spectral decomposition level in the decomposition coefficient set as the target decomposition coefficient, and sequentially extract the level noise feature and the level signal feature from the target spectral decomposition level;

[0132] Fuse the level noise feature and the level signal feature into a level feature, and generate a level threshold according to the level feature;

[0133] Judge whether the level threshold is less than the target decomposition coefficient;

[0134] Otherwise, return to the step of selecting the spectral decomposition levels in the spectral decomposition level set one by one as the target spectral decomposition levels;

[0135] If so, add the target spectral decomposition level to a preset reconstruction level set. When the target spectral decomposition level is the last level in the spectral decomposition level set, a standard reconstruction level set is obtained;

[0136] Use the following multi-level reconstruction algorithm to reconstruct the standard reconstruction level set to obtain a target denoised spectral image:

[0137]

[0138] where x j+1,k refers to the k-th decomposition coefficient of the (j + 1)-th spectral decomposition level in the spectral decomposition level set, s j refers to the scale coefficient of the j-th spectral decomposition level, n refers to the signal scale serial number of the target oat spectral image, N refers to the signal scale size of the target oat spectral image, h k-2n refers to the first filtering coefficient of the target oat spectral image at the (k - 2n)-th signal scale, d j refers to the detail coefficient of the j-th spectral decomposition level, g k-2n refers to the second filtering coefficient of the target oat spectral image at the (k - 2n)-th signal scale.

[0139] In the embodiments of the present invention, by using the multi-level decomposition algorithm to perform filtering decomposition on the target oat spectral image, a spectral decomposition level set and a corresponding decomposition coefficient set of the spectral decomposition level set are obtained, and the standard reconstruction level set is reconstructed by using the following multi-level reconstruction algorithm to obtain a target denoised spectral image, which can separate the noise signal from the original signal, thus facilitating the filtering of noise.

[0140] Specifically, the sequentially extracting the hierarchical noise feature and the hierarchical signal feature from the target spectral decomposition level means selecting the levels in the target spectral decomposition level that are in the noise level interval as the hierarchical noise feature, and taking the overall signal amplitude of the target spectral decomposition level as the hierarchical signal feature.

[0141] Specifically, the weighted fusion algorithm can be used to fuse the hierarchical noise feature and the hierarchical signal feature into a hierarchical feature, and algorithms such as the soft threshold algorithm or the hard threshold algorithm can be used to generate a hierarchical threshold according to the hierarchical feature.

[0142] In the embodiments of the present invention, by successively performing image denoising, wavelength calibration, and image enhancement operations on the target oat spectral image, a target standard spectral image is obtained, which can make the target oat spectral image clearer, with more obvious details, and at the same time retain more spectral details, thereby enhancing the coverage range of subsequent target spectral features.

[0143] S4. Vectorize the target component parameter group into target component features, downsample the target standard spectral image into target spectral features according to the target component features, fuse the target component features and the target spectral features into target oat features, and use all the target oat features and the antioxidant data set to train a preset antioxidant activity model to obtain an antioxidant analysis model.

[0144] In the embodiments of the present invention, the vectorization of the target component parameter group into target component features refers to performing a normalization operation on each component parameter in the target component parameter group to obtain a target standard parameter group, and vectorizing the standard parameters in the target standard parameter group one by one and then splicing them into target component features.

[0145] In the embodiments of the present invention, referring to Figure 3 as shown, the downsampling of the target standard spectral image into target spectral features according to the target component features includes:

[0146] S31. Extract the total dimension of the feature vectors from the target component features;

[0147] S32. Use a multi-level convolutional layer to extract various spectral spatial features from the target standard spectral image;

[0148] S33. Perform a frequency domain transformation on the target standard spectral image to obtain a target spectral frequency domain, and extract various spectral frequency domain features from the target spectral frequency domain;

[0149] S34. Aggregate all the spectral spatial features and all the spectral frequency domain features into a primary spectral feature set;

[0150] S35. Use the total dimension of the feature vectors as the number of clustering clusters to perform a clustering operation on the primary spectral feature set to obtain a spectral feature class set;

[0151] S36. Perform a downsampling operation on each spectral feature class in the spectral feature class set to obtain multiple weighted spectral center features, and aggregate all the weighted spectral center features into target spectral features.

[0152] Specifically, the total dimension of the feature vectors refers to the total number of component features of the target component features. For example, when the target component features are composed of the feature vectors of the total polysaccharide content, the total flavonoid content, the reducing sugar content, and the total phenol content, the total dimension of the feature vectors is 4.

[0153] Specifically, the spectral spatial features may be features such as the texture features, shape features, and edge features of the target standard spectral image, and the spectral frequency domain features may be features such as the power spectrum, frequency distribution, and absorption peak of the target standard spectral image.

[0154] Specifically, the fast Fourier transform algorithm can be used to perform frequency domain conversion on the target standard spectral image to obtain the target spectral frequency domain. The K-means algorithm or the Density-Based Spatial Clustering of Applications with Noise (DBSCAN) algorithm can be used to perform clustering operations on the primary spectral feature set with the total dimension of the feature vectors as the number of clustering clusters to obtain the spectral feature class set.

[0155] Specifically, the fusion of the target component features and the target spectral features into the target oat features includes:

[0156] Performing a global pooling operation on the target component features to obtain the dimension-reduced component features;

[0157] Performing a global pooling operation on the target spectral features to obtain the dimension-reduced spectral features;

[0158] Using the following multi-dimensional oat feature fusion algorithm to perform feature fusion on the dimension-reduced component features and the dimension-reduced spectral features to obtain the target oat features:

[0159]

[0160] where Z refers to the target oat features, i refers to the feature dimension serial number, E refers to the total dimension of the feature vectors of the dimension-reduced component features, and the total dimension of the feature vectors of the dimension-reduced component features is the same as the total dimension of the feature vectors of the dimension-reduced spectral features. Softmax is a normalization function, Q i is the feature vector of the i-th dimension feature in the dimension-reduced component features, α, β, and γ are the fusion coefficient matrices preset by the multi-dimensional oat feature fusion algorithm, w() is a dimension function, refers to Q i the dimension of the β vector, refers to R i the dimension of the β vector, and T is the transpose symbol.

[0161] Specifically, the global pooling operation on the target component features to obtain the dimension-reduced component features refers to performing a normalization dimension reduction operation on the features of each dimension of the target component features, and the fusion coefficient matrix is the fusion coefficient matrix of the self-attention mechanism.

[0162] Specifically, by using the multi-dimensional oat feature fusion algorithm to perform feature fusion on the dimension-reduced component features and the dimension-reduced spectral features, target oat features can be obtained. The self-attention mechanism can be used to achieve feature fusion, which can achieve feature fusion while retaining the structural correlation between features, and improve the representation degree of the target oat features.

[0163] Specifically, the antioxidant activity model can be a model such as a convolutional neural network, a self-attention model, or a deep residual network, where the input is oat features and the output is antioxidant data.

[0164] In the embodiment of the present invention, training the preset antioxidant activity model by using all the target oat features and the antioxidant data set to obtain an antioxidant analysis model includes:

[0165] Analyzing the primary antioxidant data corresponding to the target oat features one by one by using the preset antioxidant activity model, and aggregating all the primary antioxidant data into a primary antioxidant data set;

[0166] Calculating the loss value of the antioxidant activity model according to the primary antioxidant data set and the antioxidant data set;

[0167] Judging whether the loss value is greater than a preset loss value threshold;

[0168] If so, updating the model parameters of the antioxidant activity model according to the loss value, and returning to the step of analyzing the primary antioxidant data corresponding to the target oat features one by one by using the preset antioxidant activity model;

[0169] If not, using the updated antioxidant activity model as the antioxidant analysis model.

[0170] Specifically, a global loss value function such as a squared loss value or a logarithmic loss value can be used to calculate the loss value of the antioxidant activity model according to the primary antioxidant data set and the antioxidant data set, and a backpropagation algorithm such as a batch gradient descent algorithm or a full gradient descent algorithm can be used to update the model parameters of the antioxidant activity model according to the loss value.

[0171] In the embodiments of the present invention, by downsampling the target standard spectral image into a target spectral feature according to the target component feature, it can be ensured that the feature dimensions of the target component feature and the target spectral feature are the same, which is convenient for subsequent feature fusion according to the structure. By fusing the target component feature and the target spectral feature into a target oat feature, the dimension of the feature can be reduced, the magnitude of the calculation can be reduced, and the calculation efficiency can be improved. By using all the target oat features and the antioxidant data set to train a preset antioxidant activity model, an antioxidant analysis model can be obtained, which can improve the accuracy of antioxidant activity analysis.

[0172] S5. Hydrolyze the to-be-detected oat seeds into to-be-detected hydrolyzates, respectively generate the to-be-detected spectral images and to-be-detected component parameter groups of the to-be-detected hydrolyzates, generate to-be-detected oat features according to the to-be-detected spectral images and the to-be-detected component parameter groups, analyze the analysis antioxidant data of the to-be-detected oat features by using the antioxidant analysis model, and use the analysis antioxidant data as the antioxidant activity of the to-be-detected oat seeds.

[0173] In the embodiments of the present invention, the method of hydrolyzing the to-be-detected oat seeds into to-be-detected hydrolyzates is the same as the method in the above step S1 of crushing multiple oat seeds of different types into an oat flour set and performing cellulose hydrolysis on each oat flour in the oat flour set by using cellulase to obtain an oat hydrolyzate set, which will not be elaborated here.

[0174] Specifically, the step of respectively generating the to-be-detected spectral images and to-be-detected component parameter groups of the to-be-detected hydrolyzates includes generating the to-be-detected spectral images of the to-be-detected hydrolyzates and generating the to-be-detected component parameter groups of the to-be-detected hydrolyzates. Among them, the method of generating the to-be-detected spectral images of the to-be-detected hydrolyzates is the same as the method in the above step S1 of performing spectral imaging on the oat hydrolyzate set to obtain an oat spectral atlas, and the method of generating the to-be-detected component parameter groups of the to-be-detected hydrolyzates is the same as the method in the above step S2 of performing keto-phenol detection on each oat hydrolyzate in the oat hydrolyzate set to obtain an oat component parameter group set, which will not be elaborated here.

[0175] Specifically, the method for generating the to-be-detected oat characteristics based on the to-be-detected spectral image and the to-be-detected component parameter group is the same as the method in the above step S3 of successively selecting the oat spectral images in the oat spectral atlas as the target oat spectral images, selecting the oat component parameter group corresponding to the target oat spectral image in the oat component parameter group set as the target component parameter group, and successively performing image denoising, wavelength calibration, and image enhancement operations on the target oat spectral image to obtain the target standard spectral image, and the method in the above step S4 of vectorizing the target component parameter group into target component characteristics, downsampling the target standard spectral image into target spectral characteristics according to the target component characteristics, and fusing the target component characteristics and the target spectral characteristics into target oat characteristics, which will not be elaborated here.

[0176] In the embodiment of the present invention, by using the antioxidant analysis model to analyze the antioxidant data of the to-be-detected oat characteristics, and taking the analyzed antioxidant data as the antioxidant activity of the to-be-detected oat species, the antioxidant data can be analyzed according to the spectral characteristics after cellulase treatment of the to-be-detected oat species and the characteristics of substances such as glycone phenols in combination with a deep learning algorithm, improving the efficiency of antioxidant data analysis.

[0177] In the embodiment of the present invention, a plurality of oat species of different types are crushed into an oat flour set, and cellulase is used to perform cellulose hydrolysis on each oat flour in the oat flour set to obtain an oat hydrolyzate set. The cellulase can be used to dissolve the cellulose of the oat species, thereby increasing the antioxidant activity value of the oats, making the experimental data clearer and more convenient to observe. By performing spectral imaging on the oat hydrolyzate set, an oat spectral atlas can be obtained, and the material imaging of each oat hydrolyzate in the oat hydrolyzate set can be obtained, thereby obtaining more accurate measurement data of the antioxidant activity. By performing glycone phenol detection on each oat hydrolyzate in the oat hydrolyzate set, an oat component parameter group set can be obtained, and parameter information of substances such as carbohydrates, ketones, and phenols related to antioxidant can be obtained. By using the free radical scavenging method to perform antioxidant detection on the oat hydrolyzate set, an antioxidant data set can be obtained, and the true measured antioxidant data set corresponding to the oat hydrolyzate set can be analyzed according to a unified scale, facilitating subsequent model training. By successively performing image denoising, wavelength calibration, and image enhancement operations on the target oat spectral image, a target standard spectral image can be obtained, making the target oat spectral image clearer, with more obvious details, and retaining more spectral details at the same time, thereby enhancing the coverage range of the subsequent target spectral characteristics.

[0178] By downsampling the target standard spectral image into target spectral features according to the target component features, it can be ensured that the feature dimensions of the target component features and the target spectral features are the same, facilitating subsequent feature fusion according to the structure. By fusing the target component features and the target spectral features into target oat features, the dimension of the features can be reduced, the order of magnitude of the calculation can be reduced, and the calculation efficiency can be improved. By using all the target oat features and the antioxidant data set to train a preset antioxidant activity model, an antioxidant analysis model can be obtained, improving the accuracy of antioxidant activity analysis. By using the antioxidant analysis model to analyze the antioxidant data of the to-be-tested oat features and taking the analyzed antioxidant data as the antioxidant activity of the to-be-tested oat species, the antioxidant data can be analyzed based on the spectral features after cellulase treatment of the to-be-tested oat species and the features of substances such as glycone phenol in combination with deep learning algorithms, improving the efficiency of antioxidant data analysis. Therefore, the method for detecting the antioxidant activity of oat species based on cellulase proposed by the present invention can solve the problem of low efficiency in detecting the antioxidant activity of oat species.

[0179] As Figure 4 shown, it is a functional module diagram of a device for detecting the antioxidant activity of oat species based on cellulase provided by an embodiment of the present invention.

[0180] The device 100 for detecting the antioxidant activity of oat species based on cellulase according to the present invention can be installed in an electronic device. According to the functions achieved, the device 100 for detecting the antioxidant activity of oat species based on cellulase can include a spectral imaging module 101, a parameter detection module 102, an image enhancement module 103, a model training module 104, and an activity detection module 105. The modules in the present invention can also be referred to as units, which refer to a series of computer program segments that can be executed by a processor of an electronic device and can complete fixed functions, and are stored in the memory of the electronic device.

[0181] In this embodiment, the functions of each module / unit are as follows:

[0182] The spectral imaging module 101 is used to crush multiple oat species of different types into an oat flour set, perform cellulose hydrolysis on each oat flour in the oat flour set using cellulase to obtain an oat hydrolysate set, and perform spectral imaging on the oat hydrolysate set to obtain an oat spectral atlas set;

[0183] The parameter detection module 102 is used to perform glycone phenol detection on each oat hydrolysate in the oat hydrolysate set to obtain an oat component parameter group set, and perform antioxidant detection on the oat hydrolysate set using the free radical scavenging method to obtain an antioxidant data set;

[0184] The image enhancement module 103 is configured to sequentially select the oat spectral pictures in the oat spectral atlas as target oat spectral pictures, select the oat component parameter group corresponding to the target oat spectral picture in the oat component parameter group set as the target component parameter group, and perform image denoising, wavelength calibration, and image enhancement operations on the target oat spectral pictures in sequence to obtain target standard spectral pictures;

[0185] The model training module 104 is configured to vectorize the target component parameter group into target component features, downsample the target standard spectral pictures into target spectral features according to the target component features, fuse the target component features and the target spectral features into target oat features, and use all the target oat features and the antioxidant data set to train a preset antioxidant activity model to obtain an antioxidant analysis model. Wherein, the fusing of the target component features and the target spectral features into target oat features includes: performing global pooling operation on the target component features to obtain dimension-reduced component features; performing global pooling operation on the target spectral features to obtain dimension-reduced spectral features; using the following multi-dimensional oat feature fusion algorithm to perform feature fusion on the dimension-reduced component features and the dimension-reduced spectral features to obtain target oat features:

[0186]

[0187] Wherein, Z refers to the target oat feature, i refers to the feature dimension serial number, E refers to the total dimension of the feature vectors of the dimension-reduced component features, and the total dimension of the feature vectors of the dimension-reduced component features is the same as the total dimension of the feature vectors of the dimension-reduced spectral features, softmax is a normalization function, Q i is the feature vector of the i-th dimension feature in the dimension-reduced component features, α, β, γ are the preset fusion coefficient matrices of the multi-dimensional oat feature fusion algorithm, w() is a dimension function, refers to Q i the dimension of the β vector, refers to R i the dimension of the β vector, T is the transpose symbol;

[0188] The activity detection module 105 is configured to hydrolyze the to-be-detected oat seeds into to-be-detected hydrolyzates, respectively generate to-be-detected spectral pictures and to-be-detected component parameter groups of the to-be-detected hydrolyzates, generate to-be-detected oat features according to the to-be-detected spectral pictures and the to-be-detected component parameter groups, analyze the antioxidant data of the to-be-detected oat features by using the antioxidant analysis model, and use the analyzed antioxidant data as the antioxidant activity of the to-be-detected oat seeds.

[0189] Specifically, each module in the antioxidant activity detection device 100 for oat seeds based on cellulase in the embodiments of the present invention adopts the same as the above when in useFigures 1 to 3 The same technical means as the method for detecting the antioxidant activity of oat seeds based on cellulase described in [reference] can be used, and the same technical effects can be achieved, which will not be elaborated here.

[0190] In several embodiments provided by the present invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the modules is only a logical function division, and there can be other division methods in actual implementation.

[0191] The modules described as separate components may or may not be physically separated. The components shown as modules may or may not be physical units, that is, they can be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0192] In addition, in each embodiment of the present invention, the functional modules can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above integrated unit can be implemented in the form of hardware or in the form of a combination of hardware and software functional modules.

[0193] For those skilled in the art, it is obvious that the present invention is not limited to the details of the above exemplary embodiments, and without departing from the spirit or basic characteristics of the present invention, the present invention can be implemented in other specific forms.

[0194] Therefore, from any point of view, the embodiments should be regarded as exemplary and non-restrictive. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, all changes falling within the meaning and scope of the equivalent elements of the claims are intended to be included in the present invention. Any reference signs attached to the claims should not be regarded as limiting the claimed rights.

[0195] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Among them, artificial intelligence (AI) uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use knowledge to obtain the best results in theory, methods, technologies, and application systems.

[0196] In addition, obviously, the word "including" does not exclude other units or steps, and the singular does not exclude the plural. The multiple units or devices described in the system embodiments can also be implemented by one unit or device through software or hardware. The words such as first and second are used to represent names and do not represent any specific order.

[0197] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for detecting the antioxidant activity of oat seeds based on cellulase, characterized in that, The method includes: S1: Crushing multiple oat seeds of different varieties into an oat flour set, subjecting each oat flour in the oat flour set to cellulose hydrolysis using cellulase to obtain an oat hydrolyzate set, performing spectral imaging on the oat hydrolyzate set to obtain an oat spectral atlas set; S2: Detecting keto-phenols in each oat hydrolyzate in the oat hydrolyzate set to obtain an oat component parameter group set, performing antioxidant detection on the oat hydrolyzate set using the free radical scavenging method to obtain an antioxidant data set; S3: Sequentially selecting an oat spectral image in the oat spectral atlas set as a target oat spectral image, selecting the oat component parameter group corresponding to the target oat spectral image in the oat component parameter group set as a target component parameter group, and sequentially performing image denoising, wavelength calibration, and image enhancement operations on the target oat spectral image to obtain a target standard spectral image; S4: Vectorizing the target component parameter group into target component features, downsampling the target standard spectral image into target spectral features according to the target component features, fusing the target component features and the target spectral features into target oat features, and training a preset antioxidant activity model using all the target oat features and the antioxidant data set to obtain an antioxidant analysis model, wherein the fusing of the target component features and the target spectral features into target oat features includes: S41: Performing global pooling operation on the target component features to obtain dimension-reduced component features; S42: Performing global pooling operation on the target spectral features to obtain dimension-reduced spectral features; S43: Using the following multi-dimensional oat feature fusion algorithm to perform feature fusion on the dimension-reduced component features and the dimension-reduced spectral features to obtain target oat features: Among them, Z refers to the target oat feature, i refers to the serial number of the feature dimension, E refers to the total dimension of the feature vectors of the dimension-reduced component features, and the total dimension of the feature vectors of the dimension-reduced component features is the same as the total dimension of the feature vectors of the dimension-reduced spectral features. softmax is a normalization function, Q i is the feature vector of the i-th dimension feature in the dimension-reduced component features, α, β, and γ are the preset fusion coefficient matrices of the multi-dimensional oat feature fusion algorithm, and w() is a dimension function, refers to Q i the dimension of the β vector, refers to R i the dimension of the β vector, and T is the transpose symbol; S5: Hydrolyzing the oat seeds to be tested into test hydrolyzates, respectively generating test spectral images and test component parameter groups of the test hydrolyzates, generating test oat features according to the test spectral images and the test component parameter groups, analyzing the analysis antioxidant data of the test oat features using the antioxidant analysis model, and taking the analysis antioxidant data as the antioxidant activity of the oat seeds to be tested.

2. The antioxidant activity detection method of oat seeds based on cellulase according to claim 1, characterized in that, The crushing of multiple oat seeds of different varieties into an oat flour set includes: Aggregating multiple oat seeds of different varieties into an oat seed set, and sequentially selecting one oat seed in the oat seed set as a target oat seed; Sequentially performing cleaning and drying operations on the target oat seed to obtain a target clean oat seed; Sequentially performing crushing and sieving operations on the target clean oat seed to obtain a target primary oat flour; Screening out oat flour of a preset quality from the target primary oat flour as target oat flour, and aggregating all the target oat flour into an oat flour set.

3. The antioxidant activity detection method of oat seeds based on cellulase according to claim 1, characterized in that, The use of cellulase to perform cellulose hydrolysis on each oat flour in the oat flour set to obtain an oat hydrolyzate set includes: Sequentially selecting an oat flour in the oat flour set as a target oat flour, stirring and dissolving the target oat flour to obtain an oat flour solution; filtering out a standard oat liquid from the oat flour solution, and performing a fiber hydrolysis reaction on the standard oat liquid using cellulase to obtain a hydrolysis reaction liquid; A primary hydrolyzate is filtered and precipitated from the hydrolysis reaction liquid, and the primary hydrolyzate is dried to obtain oat hydrolyzate, and all the oat hydrolyzates are collected into an oat hydrolyzate set.

4. The antioxidant activity detection method of oat seeds based on cellulase according to claim 1, characterized in that, The sugar, ketone and phenol detection is performed on each oat hydrolyzate in the oat hydrolyzate set to obtain an oat component parameter set, including: selecting oat hydrolysates from the oat hydrolysate collection one by one as target oat hydrolysates, and detecting the total polysaccharide content from the target oat hydrolysates using a precipitation acid reaction method; detecting the total flavonoid content from the target oat hydrolyzate using an aluminum salt colorimetric method; detecting the reducing sugar content from the target oat hydrolyzate using the phenol-sulfuric acid method; detecting the total phenol content from the target oat hydrolyzate using a sodium salt colorimetric method; The total polysaccharide content, the total flavonoid content, the reducing sugar content, and the total phenol content are aggregated into an oat component parameter group of the target oat hydrolyzate, and all oat component parameter groups are aggregated into an oat component parameter group set.

5. The antioxidant activity detection method of oat seeds based on cellulase according to claim 1, characterized in that The antioxidant test of the oat hydrolyzate set using the free radical scavenging method is performed to obtain an antioxidant data set, including: selecting oat hydrolysates from the oat hydrolysate set one by one as target oat hydrolysates, and dissolving the target oat hydrolysates into target oat solvent groups with different concentrations; using a pre-prepared free radical solution to react with each target oat solvent in the target oat solvent group to obtain a target oat reaction solvent group; Detecting the target absorbance groups corresponding to the target oatmeal reaction solvent groups using a spectrophotometer; Calculating the free radical scavenging rate of the target oat hydrolyzate based on the target absorbance group and a preset positive control group; The free radical scavenging rate is used as the antioxidant data of the target oat hydrolyzate, and all antioxidant data are compiled into an antioxidant data set.

6. The antioxidant activity detection method of oat seeds based on cellulase according to claim 1, wherein The step of sequentially performing image denoising, wavelength calibration, and image enhancement operations on the target oat spectrum image to obtain a target standard spectrum image includes: Performing threshold denoising on the target oat spectral image using a decomposition and reconstruction method to obtain a target denoised spectral image; Performing baseline correction on the target denoised spectrum image to obtain a target baseline spectrum image; Performing wavelength calibration on the target baseline spectrum image according to the pre-acquired sample spectrum image to obtain a target calibration spectrum image; A grayscale histogram of the target calibration spectrum picture is generated, and a grayscale equalization operation is performed on the target calibration spectrum picture according to the grayscale histogram to obtain a target standard spectrum picture.

7. The antioxidant activity detection method of oat seeds based on cellulase according to claim 6, wherein The method of performing threshold denoising on the target oat spectral image by using the decomposition and reconstruction method to obtain the target denoised spectral image includes: The target oat spectral image is filtered and decomposed using the following multi-layer decomposition algorithm to obtain a spectral decomposition level set and a decomposition coefficient set corresponding to the spectral decomposition level set: where x j,k refers to the k-th decomposition coefficient of the j-th spectral decomposition level in the spectral decomposition hierarchy set, s j refers to the scale coefficient of the j-th spectral decomposition level, n refers to the signal scale serial number of the target oat spectral image, N refers to the signal scale size of the target oat spectral image, h n refers to the first filtering coefficient of the target oat spectral image at the n-th signal scale, d j refers to the detail coefficient of the j-th spectral decomposition level, g n refers to the second filtering coefficient of the target oat spectral image at the n-th signal scale; Select the spectral decomposition levels in the spectral decomposition level set one by one as the target spectral decomposition levels, take the decomposition coefficients corresponding to the target spectral decomposition levels in the decomposition coefficient set as the target decomposition coefficients, and sequentially extract the level noise features and level signal features from the target spectral decomposition levels; Fuse the level noise features and the level signal features into level features, and generate a level threshold according to the level features; Determine whether the level threshold is less than the target decomposition coefficient; If not, return to the step of selecting the spectral decomposition levels in the spectral decomposition level set one by one as the target spectral decomposition levels; If so, add the target spectral decomposition levels to a preset reconstruction level set. When the target spectral decomposition level is the last level in the spectral decomposition level set, obtain a standard reconstruction level set; Reconstruct the standard reconstruction level set using the following multi-level reconstruction algorithm to obtain a target denoised spectral image: Among them, x j+1,k refers to the k-th decomposition coefficient of the (j + 1)-th spectral decomposition level in the spectral decomposition hierarchy set, s j refers to the scale coefficient of the j-th spectral decomposition level, n refers to the signal scale serial number of the target oat spectral image, N refers to the signal scale size of the target oat spectral image, h k-2n refers to the first filtering coefficient of the target oat spectral image at the (k - 2n)-th signal scale, d j refers to the detail coefficient of the j-th spectral decomposition level, g k-2n refers to the second filtering coefficient of the target oat spectral image at the (k - 2n)-th signal scale.

8. The antioxidant activity detection method of oat seeds based on cellulase according to claim 1, wherein The downsampling the target standard spectral image into a target spectral feature according to the target component feature includes: Extract the total dimension of the feature vectors from the target component feature; Use multiple convolutional layers to extract various spectral spatial features from the target standard spectral image; Perform a frequency domain transformation on the target standard spectral image to obtain a target spectral frequency domain, and extract various spectral frequency domain features from the target spectral frequency domain; Integrate all the spectral spatial features and all the spectral frequency domain features into a primary spectral feature set; Use the total dimension of the feature vectors as the number of clustering clusters to perform a clustering operation on the primary spectral feature set to obtain a spectral feature class set; Perform a downsampling operation on each spectral feature class in the spectral feature class set to obtain multiple weighted spectral center features, and integrate all the weighted spectral center features into a target spectral feature.

9. The antioxidant activity detection method of oat seeds based on cellulase according to claim 1, wherein The training the preset antioxidant activity model using all the target oat features and the antioxidant data set to obtain an antioxidant analysis model includes: Analyze the primary antioxidant data corresponding to the target oat features one by one using the preset antioxidant activity model, and integrate all the primary antioxidant data into a primary antioxidant data set; Calculate the loss value of the antioxidant activity model according to the primary antioxidant data set and the antioxidant data set; Determine whether the loss value is greater than a preset loss value threshold; If so, update the model parameters of the antioxidant activity model according to the loss value, and return to the step of analyzing the primary antioxidant data corresponding to the target oat features one by one using the preset antioxidant activity model; If not, use the updated antioxidant activity model as the antioxidant analysis model.

10. An antioxidant activity detection device for oat seeds based on cellulase, characterized in that, The device includes: A spectral imaging module, configured to crush multiple oat seeds of different types into an oat flour set, perform cellulose hydrolysis on each oat flour in the oat flour set using cellulase to obtain an oat hydrolyzate set, and perform spectral imaging on the oat hydrolyzate set to obtain an oat spectral atlas set; A parameter detection module, which is used to perform keto-phenol detection on each oat hydrolyzate in the oat hydrolyzate set to obtain an oat component parameter set, and perform antioxidant detection on the oat hydrolyzate set by the free radical scavenging method to obtain an antioxidant data set; An image enhancement module, which is used to select each oat spectral image in the oat spectral atlas as a target oat spectral image one by one, select the oat component parameter set corresponding to the target oat spectral image in the oat component parameter set as the target component parameter set, and perform image denoising, wavelength calibration and image enhancement operations on the target oat spectral image in sequence to obtain a target standard spectral image; A model training module, which is used to vectorize the target component parameter set into target component features, downsample the target standard spectral image into target spectral features according to the target component features, fuse the target component features and the target spectral features into target oat features, and use all the target oat features and the antioxidant data set to train a preset antioxidant activity model to obtain an antioxidant analysis model. Among them, the fusing of the target component features and the target spectral features into target oat features includes: performing a global pooling operation on the target component features to obtain dimension-reduced component features; performing a global pooling operation on the target spectral features to obtain dimension-reduced spectral features; using the following multi-dimensional oat feature fusion algorithm to perform feature fusion on the dimension-reduced component features and the dimension-reduced spectral features to obtain target oat features: Among them, Z refers to the target oat feature, i refers to the serial number of the feature dimension, E refers to the total dimension of the feature vectors of the dimensionality-reduced component features, and the total dimension of the feature vectors of the dimensionality-reduced component features is the same as that of the feature vectors of the dimensionality-reduced spectral features. softmax is a normalization function, Q i is the feature vector of the i-th dimension feature in the dimensionality-reduced component features, α, β, and γ are the preset fusion coefficient matrices of the multi-dimensional oat feature fusion algorithm, and w() is a dimension function, refers to Q i the dimension of the β vector, refers to R i the dimension of the β vector, and T is the transpose symbol; An activity detection module, which is used to hydrolyze the to-be-detected oat seeds into to-be-detected hydrolyzates, respectively generate to-be-detected spectral images and to-be-detected component parameter sets of the to-be-detected hydrolyzates, generate to-be-detected oat features according to the to-be-detected spectral images and the to-be-detected component parameter sets, analyze the analysis antioxidant data of the to-be-detected oat features by using the antioxidant analysis model, and use the analysis antioxidant data as the antioxidant activity of the to-be-detected oat seeds.

Images