Tea leaf detection device and method and computer storage medium

Abstract

The invention is suitable for the field of tea detection, and provides a tea detection device and method and a computer storage medium, the method comprises the following steps: collecting tea multi-modal information, the multi-modal information comprising image source information, spectrum source information and aroma source information; respectively extracting unit feature information in the tea multi-modal information; constructing a feature fusion quantization formula, and calculating and generating a fusion feature value based on a plurality of pieces of unit feature information; on the basis of the fusion feature value, identifying in-rule tea leaf types, and generating a rejection control instruction on the basis of the in-rule tea leaf types; obtaining the generation frequency of the elimination control instruction in the target time period, and generating a termination control instruction; according to the method and the system, three types of information including images, spectrums and aromas are subjected to collaborative analysis, the shortages of a single detection means are effectively avoided, and the characteristics of the tea leaves are accurately analyzed from multiple dimensions. By means of quantitative analysis and intelligent identification technologies, various types of tea leaves can be distinguished sensitively and accurately, and an automatic removing process is matched.

Description

Technical Field

[0001] The invention belongs to the field of tea detection, and in particular relates to a tea detection device, method and computer storage medium. Background Art

[0002] In the industrial production and bagging process of tea, it is crucial to ensure the consistency of the tea category in each bag. However, due to various factors, the mixing of different types of tea frequently occurs. The traditional tea picking process mostly relies on manual labor, and it is difficult to ensure that the tea picked each time is of the same variety, and mixing with other varieties of tea often occurs. Moreover, in different stages of tea processing, such as withering, rolling, fermentation, etc., if the operation is improper or the equipment is not thoroughly cleaned, cross-contamination of different types of tea will also occur. Different types of tea differ in appearance, flavor, chemical composition, etc. Once mixed, it will seriously affect the quality and market value of the tea. Consumers expect to buy a single type of tea with specific flavor and quality characteristics. Mixed tea cannot meet this demand, which reduces consumer satisfaction and damages the brand image of tea production companies.

[0003] At present, although there are some manual sorting methods, they are inefficient and highly subjective, and it is difficult to meet the precision and efficiency requirements of large-scale industrial production. Some existing automated detection technologies, such as simple image recognition, are difficult to accurately distinguish different types of tea due to the high similarity in the appearance of tea leaves, resulting in poor results in removing mixed tea leaves. Therefore, the development of an efficient and accurate tea detection device, method and computer storage medium that can effectively identify the main tea leaves and remove non-main tea leaves is of great practical significance for improving the quality and efficiency of tea production. Summary of the invention

[0004] The object of the present invention is to provide a tea detection device, method and computer storage medium, aiming to solve the problems raised in the above background technology.

[0005] The present invention is implemented in this way. On the one hand, a tea detection method comprises: Collecting tea multimodal information, wherein the multimodal information includes image source information, spectrum source information and aroma source information; Extract unit feature information from tea multimodal information respectively; Based on several unit feature information, a fusion feature value is calculated and generated; Based on the fused feature values, the types of tea within the specifications are identified, and a rejection control instruction is generated based on the types of tea within the specifications; Obtain the number of times the elimination control instruction is generated within the target period, and generate a termination control instruction.

[0006] As a further solution of the present invention, the multimodal information includes image source information, spectrum source information and aroma source information, specifically including: Collect several images of tea leaves to be tested at different angles; Collecting spectral data of the tea leaves to be tested within a preset wavelength range; Collect electrical signal data with aroma characteristics.

[0007] As a further solution of the present invention, the extracting unit feature information from the multimodal information of tea leaves specifically includes: Based on the images of tea leaves to be tested at different angles and the preset convolutional neural network architecture learning model, the connection relationship between the edge pixels of the tea leaves is determined; Based on the connection relationship of the edge pixels of the tea leaves, the ratio of the perimeter and area of ​​the tea leaves to be detected is calculated and generated. ; Extract the texture complexity features of the tea leaves to be tested from different angles, and generate the entropy value of the gray-level co-occurrence matrix of the tea leaves to be tested based on the gray-level co-occurrence matrix algorithm. ; Based on the spectral data of the tea leaves to be tested in the preset wavelength range, the average slope of the preset wavelength range is calculated. ; Calculate the relative content of the targeted aroma component in the total aroma components based on the electrical signal data with aroma characteristics .

[0008] As a further solution of the present invention, the step of calculating and generating a fusion feature value based on a plurality of unit feature information specifically includes: Determine the ratio of the perimeter to the area of ​​the tea leaves to be tested , Entropy value of the gray-level symbiosis matrix of the tea leaves to be tested , the average slope of the preset wavelength range and the relative content of targeted aroma components in the total aroma components The corresponding weight coefficient and ; Calculate and generate fusion eigenvalues ; The fusion feature value The calculation process is: .

[0009] As a further solution of the present invention, the method of identifying the type of tea leaves within the specification based on the fusion feature value and generating a rejection control instruction based on the type of tea leaves within the specification specifically includes: Determine the fusion feature value Is it greater than the characteristic value threshold? If the fusion eigenvalues If it is greater than the characteristic value threshold, a rejection control instruction is generated; The rejection control instruction includes position information of tea leaves to be detected, and the position information of tea leaves to be detected is determined based on the coordinate positioning system of the image source information acquisition device and the operating parameters of the tea leaf conveying device.

[0010] As a further solution of the present invention, on the other hand, a tea detection device is provided, the device comprising: A collection module, used to collect multimodal information of tea leaves; The multimodal information includes image source information, spectrum source information and aroma source information; An extraction module, used to extract unit feature information from the multimodal information of tea leaves; A calculation and generation module, used for calculating and generating a fusion feature value based on a plurality of unit feature information; An identification module, used for identifying the type of tea leaves within the specification based on the fused feature value, and generating a rejection control instruction based on the type of tea leaves within the specification; An acquisition module is used to acquire the number of times the rejection control instruction is generated within a target period; A generation module is used to generate a termination control instruction.

[0011] As a further solution of the present invention, the acquisition module specifically includes: An image acquisition unit, used to acquire a plurality of images of tea leaves to be tested at different angles; A spectrum acquisition unit, used to collect spectrum data of the tea leaves to be tested within a preset wavelength range; The electronic nose collection unit is used to collect electrical signal data with aroma characteristics.

[0012] As a further solution of the present invention, the extraction module specifically includes: A determination unit, used to determine the connection relationship of the edge pixels of the tea leaves based on the images of the tea leaves to be detected at different angles and a preset convolutional neural network architecture learning model; A calculation and generation unit is used to calculate and generate the ratio of the perimeter and area of ​​the outline of the tea leaves to be detected based on the connection relationship of the edge pixels of the tea leaves. ; An extraction unit, used for extracting texture complexity features from images of tea leaves to be detected at different angles; The first generating unit is used to generate the entropy value of the gray-level symbiosis matrix of the tea to be detected based on the gray-level symbiosis matrix algorithm. ; The first calculation unit is used to calculate the average slope of the preset wavelength range based on the spectral data of the tea leaves to be detected in the preset wavelength range. ; The second calculation unit is used to calculate the relative content of the targeted aroma component in the total aroma components based on the electrical signal data with aroma characteristics. .

[0013] As a further solution of the present invention, the identification module specifically includes: Judgment unit, used to judge the fusion feature value Is it greater than the characteristic value threshold? The second generation unit is used to fuse the eigenvalues If the value is greater than the characteristic value threshold, a rejection control instruction is generated.

[0014] As a further solution of the present invention, a computer storage medium is provided, wherein the computer storage medium stores a computer program, and when the computer program is executed by a processor, the above-mentioned tea leaf detection steps are implemented.

[0015] This method and system analyzes three types of information, namely image, spectrum and aroma, in a coordinated manner, effectively avoiding the shortcomings of a single detection method and accurately analyzing tea characteristics from multiple dimensions. With the help of quantitative analysis and intelligent recognition technology, it can keenly and accurately distinguish various types of tea, and cooperate with the automated elimination process to effectively overcome the industry's stubborn problem of tea mixing and protect tea quality in all aspects. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a main flow chart of a tea testing method.

[0017] Figure 2 The invention discloses a flow chart of collecting multimodal information of tea leaves in a tea leaves detection method.

[0018] Figure 3 The invention discloses a flow chart of a tea detection method for identifying the type of tea within the specification based on fused characteristic values ​​and generating a rejection control instruction based on the type of tea within the specification.

[0019] Figure 4 It is a main structural diagram of a tea detection device.

[0020] Figure 5 The present invention is a structural block diagram of a collection module in a tea detection device.

[0021] Figure 6 The present invention is a structural block diagram of an extraction module in a tea detection device.

[0022] Figure 7 The present invention is a structural block diagram of an identification module in a tea detection device. DETAILED DESCRIPTION

[0023] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0024] The specific implementation of the present invention is described in detail below in conjunction with specific embodiments.

[0025] The invention provides a tea detection device, method and computer storage medium, which solve the technical problems in the background technology.

[0026] like Figure 1 FIG. 1 is a main flow chart of a tea detection method provided by an embodiment of the present invention, wherein the tea detection method comprises: Step S100: collecting tea multimodal information, wherein the multimodal information includes image source information, spectrum source information and aroma source information; Step S200: extracting unit feature information from the tea multimodal information respectively; Step S300: Calculate and generate a fusion feature value based on a plurality of unit feature information; Step S400: identifying the type of tea leaves within the specification based on the fused feature value, and generating a rejection control instruction based on the type of tea leaves within the specification; Step S500: obtaining the number of times the elimination control instruction is generated within the target period, and generating a termination control instruction; When this embodiment is applied, the multimodal information of tea leaves is first collected. The image source information is collected by high-resolution equipment, which can clearly present the subtle texture, unique shape and precise color of tea leaves, providing a solid foundation for judging the appearance characteristics of tea leaves. The spectral source information uses a spectrometer to cover a specific wavelength range, accurately reflecting the chemical composition and internal structure of tea leaves. Different tea leaves present unique absorption peaks and reflectivity changes on the spectrum due to differences in chemical composition. The aroma source information is captured by an electronic nose to perceive the aroma components emitted by tea leaves; the unit feature information extraction and the construction of feature fusion quantitative formulas realize the deep mining and integration of multimodal information of tea leaves. By extracting the diversity characteristics of image source information, spectral source information and aroma source information, the diversity characteristics are integrated into the quantitative formula according to the preset weight coefficient to generate a fusion feature value that fully reflects the characteristics of tea leaves. The fusion feature value is compared with the pre-stored feature value database covering a rich variety of tea types, accurately identifying the types of tea leaves within the specification, and effectively distinguishing different categories of tea leaves; once the mixed tea leaves outside the specification are identified, a rejection control instruction is generated to drive the automatic rejection device to reject bad tea leaves with high accuracy and high speed, thereby reducing the tediousness and inefficiency of manual sorting. At the same time, by counting the number of times the rejection control instructions are generated within the target period, when the number of times the rejection control instructions are generated within the target period exceeds the threshold, it indicates that the purity of the batch of tea does not meet the standards, and a termination control instruction is automatically generated to achieve full automation and intelligence of the detection process.

[0027] like Figure 2 As shown, as a preferred embodiment of the present invention, multimodal information of tea leaves is collected, and the multimodal information includes image source information, spectrum source information and aroma source information: Step S101: collecting a number of images of tea leaves to be tested at different angles; Step S102: collecting spectral data of the tea leaves to be tested within a preset wavelength range; Step S103: collecting electrical signal data with aroma characteristics; When this embodiment is applied, in terms of image source information acquisition, a high-resolution image acquisition device is used to shoot the tea leaves to be detected from multiple angles. By acquiring images from several different angles, a rich visual data basis is provided for subsequent feature extraction and category recognition. Spectral source information acquisition uses a spectrometer to cover a preset wavelength range, such as 200nm-1100nm. The spectrometer can accurately detect the absorbance or reflectance of tea leaves at different wavelengths. These spectral data deeply reflect the chemical composition and internal structure characteristics of tea leaves. Due to the difference in chemical composition, different types of tea leaves show unique absorption peaks and reflectance changes on the spectrum, providing a key chemical feature basis for distinguishing tea categories. Aroma source information acquisition relies on an electronic nose device equipped with a sensor array. Within a specific time period, such as 30 seconds, the electronic nose captures the aroma released by the tea leaves and collects electrical signal data with aroma characteristics. Its detection sensitivity for common tea aroma components can reach 1ppm, and it can keenly perceive the complex aroma components emitted by tea leaves. The types and contents of aroma components of different varieties of tea leaves have their own characteristics. The aroma characteristics further enrich the detection dimension and make up for the lack of image and spectral information.

[0028] As a preferred embodiment of the present invention, the extracting of unit feature information from the multimodal information of tea leaves specifically includes: Step S201: determining the connection relationship of the edge pixels of the tea leaves based on the images of the tea leaves to be detected at different angles and a preset convolutional neural network architecture learning model; Step S202: Calculate and generate the ratio of the perimeter and area of ​​the tea leaves to be detected based on the connection relationship of the tea leaf edge pixels ; Step S203: extracting texture complexity features from different angle images of the tea leaves to be detected, and generating entropy values ​​of the gray-level co-occurrence matrix of the tea leaves to be detected based on the gray-level co-occurrence matrix algorithm ; Step S204: Calculate the average slope of the preset wavelength range based on the spectral data of the tea leaves to be tested in the preset wavelength range ; Step S205: Calculate the relative content of the targeted aroma component in the total aroma components based on the electrical signal data with aroma characteristics ; When this embodiment is applied, for the image source information, it relies on the preset convolutional neural network architecture learning model to conduct in-depth analysis of images at different angles. Through model calculation, the connection relationship of the edge pixels of the tea leaves is determined, and based on this, the ratio of the perimeter and area of ​​the outline of the tea leaves to be detected is accurately calculated. At the same time, the gray-level co-occurrence matrix algorithm is used to extract the texture complexity features from the image and generate the entropy value of the gray-level co-occurrence matrix of the tea to be tested. , entropy value The size reflects the complexity of the tea texture. Based on the spectral source information, the average slope of the spectral curve within the preset wavelength range is calculated based on the collected spectral data. The slope change reflects the changing trend of the absorption characteristics of tea in this wavelength range. In the aroma source information processing, based on the collected electrical signal data, a specific algorithm is used to calculate the relative content of the targeted aroma component in the total aroma component. The relative contents of aroma components in different tea varieties vary.

[0029] As a preferred embodiment of the present invention, based on a plurality of unit feature information, the fusion feature value is calculated and generated specifically: Determine the ratio of the perimeter to the area of ​​the tea leaves to be tested , Entropy value of the gray-level symbiosis matrix of the tea leaves to be tested , the average slope of the preset wavelength range and the relative content of targeted aroma components in the total aroma components The corresponding weight coefficient and ; Calculate and generate fusion eigenvalues ; The fusion feature value The calculation process is: .

[0030] It should be understood that the weight coefficient and It can be determined based on a large number of tea sample training, and , and The sample data includes tea samples from multiple different origins and multiple different varieties. For each variety of tea, images, spectra and aroma data are collected at different growth stages and processing techniques. The number of images of each sample is no less than 10, the number of spectral data collection times is no less than 15, and the number of aroma data collection times is no less than 15.

[0031] like Figure 3 As shown, as a preferred embodiment of the present invention, the identification of the type of tea leaves within the specification based on the fusion feature value and the generation of the rejection control instruction based on the type of tea leaves within the specification specifically include: Step S401: Determine fusion feature value Is it greater than the characteristic value threshold? Step S402: If the fusion feature value If it is greater than the characteristic value threshold, a rejection control instruction is generated; The rejection control instruction includes the position information of the tea leaves to be detected, and the position information of the tea leaves to be detected is determined based on the coordinate positioning system of the image source information acquisition device and the operating parameters of the tea leaf conveying device; In the application of this embodiment, in the process of identifying the type of tea within the specification and generating the rejection control instruction, a characteristic value threshold is first set, which is determined based on a large amount of tea sample data training and actual production needs. Compare and judge with the characteristic value threshold. If it is greater than the characteristic value threshold, it indicates that the tea leaves to be tested may belong to the category of non-standard tea leaves, and a rejection control instruction is quickly generated at this time. The rejection control instruction contains key information about the position of the tea leaves to be tested, and the acquisition of this information depends on the coordinate positioning system of the image source information acquisition device and the operating parameters of the tea conveyor device. The coordinate positioning system of the image acquisition device can accurately record the relative position of the tea leaves in the field of view of the device in the captured image, and the operating parameters of the tea conveyor device, such as the conveying speed, conveying direction, and start and stop time, can be used to convert the actual position of the tea leaves on the production line. By combining and calculating the data of the two, the specific position of the tea leaves to be tested on the production line can be accurately determined, ensuring that the automated rejection device can accurately remove non-standard tea leaves on the production line with preset rejection accuracy and rejection speed according to the rejection control instruction.

[0032] like Figure 4 As shown, as another preferred embodiment of the present invention, on the other hand, a tea detection device comprises: The collection module 100 is used to collect multimodal information of tea leaves; The multimodal information includes image source information, spectrum source information and aroma source information; The extraction module 200 is used to extract the unit feature information from the multimodal information of tea leaves respectively; A calculation and generation module 300 is used to calculate and generate a fusion feature value based on a plurality of unit feature information; The identification module 400 is used to identify the type of tea leaves within the specification based on the fused feature value, and generate a rejection control instruction based on the type of tea leaves within the specification; An acquisition module 500 is used to acquire the number of times the rejection control instruction is generated within a target period; The generating module 600 is used to generate a termination control instruction.

[0033] When this embodiment is applied, the collection module 100 collects multimodal information of tea leaves, the extraction module 200 extracts unit feature information in the multimodal information of tea leaves respectively, and based on several unit feature information, the calculation generation module 300 calculates and generates a fusion feature value, based on the fusion feature value, the identification module 400 identifies the type of tea leaves within the specification, and generates a rejection control instruction based on the type of tea leaves within the specification, generates a rejection control instruction, the acquisition module 500 obtains the number of times the rejection control instruction is generated within the target time period, and the generation module 600 generates a termination control instruction.

[0034] like Figure 5 As shown, as another preferred embodiment of the present invention, the acquisition module 100 specifically includes: An image acquisition unit 101 is used to acquire a plurality of images of tea leaves to be detected at different angles; The spectrum acquisition unit 102 is used to acquire spectrum data of the tea leaves to be tested within a preset wavelength range; The electronic nose collection unit 103 is used to collect electrical signal data with aroma characteristics.

[0035] When this embodiment is used, the image acquisition unit 101 acquires images of several tea leaves to be detected at different angles, the spectrum acquisition unit 102 acquires spectrum data of the tea leaves to be detected within a preset wavelength range, and the electronic nose acquisition unit 103 acquires electrical signal data with aroma characteristics.

[0036] like Figure 6 As shown, as another preferred embodiment of the present invention, the extraction module 200 specifically includes: A determination unit 201 is used to determine the connection relationship of the edge pixels of the tea leaves based on the images of the tea leaves to be detected at different angles and a preset convolutional neural network architecture learning model; The calculation and generation unit 202 is used to calculate and generate the ratio of the perimeter and area of ​​the outline of the tea leaves to be detected based on the connection relationship of the edge pixels of the tea leaves. ; An extraction unit 203 is used to extract texture complexity features from images of tea leaves to be detected at different angles; The first generating unit 204 is used to generate the entropy value of the gray level symbiosis matrix of the tea to be detected based on the gray level symbiosis matrix algorithm. ; The first calculation unit 205 is used to calculate the average slope of the preset wavelength range based on the spectral data of the tea leaves to be detected in the preset wavelength range. ; The second calculation unit 206 is used to calculate the relative content of the targeted aroma component in the total aroma components based on the electrical signal data with aroma characteristics. .

[0037] When this embodiment is used, based on the different angle images of the tea leaves to be detected and the preset convolutional neural network architecture learning model, the determination unit 201 determines the connection relationship of the edge pixels of the tea leaves, and based on the connection relationship of the edge pixels of the tea leaves, the calculation generation unit 202 calculates and generates the ratio of the perimeter of the outline of the tea leaves to be detected to the area. The extraction unit 203 extracts the texture complexity features in the images of the tea leaves to be detected at different angles. Based on the gray-level co-occurrence matrix algorithm, the first generation unit 204 generates the entropy value of the gray-level co-occurrence matrix of the tea leaves to be detected. Based on the spectral data of the tea leaves to be detected in the preset wavelength range, the first calculation unit 205 calculates the average slope of the preset wavelength range. Based on the electrical signal data with aroma characteristics, the second calculation unit 206 calculates the relative content of the targeted aroma component in the total aroma components. .

[0038] like Figure 7 As shown, as another preferred embodiment of the present invention, the identification module 400 specifically includes: The judging unit 401 is used to judge the fusion feature value Is it greater than the characteristic value threshold? The second generating unit 402 is used to fuse the feature values If it is greater than the characteristic value threshold, a rejection control instruction is generated; The judging unit 401 judges the fusion feature value Is it greater than the eigenvalue threshold? If the fusion eigenvalue If the value is greater than the characteristic value threshold, the second generating unit 402 generates a rejection control instruction.

[0039] In another embodiment of the present invention, a computer storage medium is provided, which stores a computer program. The computer program includes program instructions. When the program instructions are executed by a processor, all or part of the process in the method of the above embodiment is implemented. The computer program can also be used to instruct related hardware to complete it. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by the processor, the steps of each of the above method embodiments can be implemented.

[0040] In the above embodiment of the present invention, a tea detection method is provided, and a tea detection device and a computer storage medium are provided. First, the multimodal information of tea is collected. The image source information is collected by high-resolution equipment, which can clearly present the subtle texture, unique shape and precise color of tea, providing a solid foundation for judging the appearance characteristics of tea. The spectral source information uses a spectrometer to cover a specific wavelength range, accurately reflecting the chemical composition and internal structure of tea. Different teas present unique absorption peaks and reflectivity changes on the spectrum due to differences in chemical composition. The aroma source information is captured by an electronic nose to perceive the aroma components emitted by tea; the unit feature information is extracted and the feature fusion quantification formula is constructed to achieve deep mining and integration of multimodal information of tea. By extracting the diversity features of image source information, spectral source information and aroma source information, the diversity features are integrated into the quantification formula according to the preset weight coefficient to generate a fusion feature value that fully reflects the characteristics of tea. The fusion feature value is compared with the pre-stored feature value database covering a rich variety of tea types to accurately identify the types of tea within the specification and effectively distinguish different types of tea; once the mixed tea outside the specification is identified, a rejection control instruction is generated to drive the automatic rejection device to reject bad tea with high precision and high speed, thereby reducing the tediousness and inefficiency of manual sorting. At the same time, by counting the number of rejection control instructions generated within the target period, when the number of rejection control instructions generated within the target period exceeds the number threshold, it indicates that the purity of the batch of tea does not meet the standard, and the termination control instruction is automatically generated to achieve full automation and intelligence of the detection process; this method and system will analyze the three types of information, image, spectrum, and aroma, effectively avoid the shortcomings of a single detection method, and accurately analyze the characteristics of tea from multiple dimensions. With the help of quantitative analysis and intelligent recognition technology, it is possible to keenly and accurately distinguish various types of tea, cooperate with the automated rejection process, effectively overcome the industry's stubborn problem of tea mixing, and protect the quality of tea in all aspects.

[0041] In order to load the above-mentioned method and system and enable it to run smoothly, the system, in addition to the various modules mentioned above, may also include more or fewer components than described above, or a combination of certain components, or different components, for example, it may include input and output devices, network access devices, buses, processors and memories, etc.

[0042] The processor may be a central processing unit, or other general-purpose processors, digital signal processors, application-specific integrated circuits, off-the-shelf programmable gate arrays or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the system, and various interfaces and lines are used to connect various parts.

[0043] The technical features of the above-described embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0044] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

[0045] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A tea detection method, characterized in that: The method comprises: Collecting tea multimodal information, wherein the multimodal information includes image source information, spectrum source information and aroma source information; Extract unit feature information from tea multimodal information respectively; Based on several unit feature information, a fusion feature value is calculated and generated; Based on the fused feature values, the types of tea within the specifications are identified, and a rejection control instruction is generated based on the types of tea within the specifications; Obtain the number of times the elimination control instruction is generated within the target period, and generate a termination control instruction; The extracting of unit characteristic information from the multimodal information of tea leaves specifically includes: Based on the images of tea leaves to be tested at different angles and the preset convolutional neural network architecture learning model, the connection relationship between the edge pixels of the tea leaves is determined; Based on the connection relationship of the edge pixels of the tea leaves, the ratio of the perimeter and area of ​​the tea leaves to be detected is calculated and generated. ; Extract the texture complexity features of the tea leaves to be tested from different angles, and generate the entropy value of the gray-level co-occurrence matrix of the tea leaves to be tested based on the gray-level co-occurrence matrix algorithm. ; Based on the spectral data of the tea leaves to be tested in the preset wavelength range, the average slope of the preset wavelength range is calculated. ; Calculate the relative content of the targeted aroma component in the total aroma components based on the electrical signal data with aroma characteristics .

2. The tea detection method according to claim 1, characterized in that: Collecting tea multimodal information, the multimodal information includes image source information, spectrum source information and aroma source information, specifically including: Collect several images of tea leaves to be tested at different angles; Collecting spectral data of the tea leaves to be tested within a preset wavelength range; Collect electrical signal data with aroma characteristics.

3. The tea detection method according to claim 1, characterized in that: The step of calculating and generating a fusion feature value based on a plurality of unit feature information specifically includes: Determine the ratio of the perimeter to the area of ​​the tea leaves to be tested , Entropy value of the gray-level symbiosis matrix of the tea leaves to be tested , the average slope of the preset wavelength range and the relative content of targeted aroma components in the total aroma components The corresponding weight coefficient and ; Calculate and generate fusion eigenvalues ; The fusion feature value The calculation process is: 。 4. The tea detection method according to claim 1, characterized in that: The method of identifying the type of tea leaves within the specification based on the fused feature value and generating a rejection control instruction based on the type of tea leaves within the specification specifically includes: Determine the fusion feature value Is it greater than the characteristic value threshold? If the fusion eigenvalues If it is greater than the characteristic value threshold, a rejection control instruction is generated; The rejection control instruction includes position information of tea leaves to be detected, and the position information of tea leaves to be detected is determined based on the coordinate positioning system of the image source information acquisition device and the operating parameters of the tea leaf conveying device.

5. A tea detection device, characterized in that: The tea detection method according to any one of claims 1 to 4 is applied, wherein the device comprises: A collection module, used to collect multimodal information of tea leaves; The multimodal information includes image source information, spectrum source information and aroma source information; An extraction module, used to extract unit feature information from the multimodal information of tea leaves; A calculation and generation module, used for calculating and generating a fusion feature value based on a plurality of unit feature information; An identification module, used to identify the type of tea leaves within the specification based on the fused feature value, and generate a rejection control instruction based on the type of tea leaves within the specification; An acquisition module is used to acquire the number of times the rejection control instruction is generated within a target period; A generation module is used to generate a termination control instruction.

6. The tea detection device according to claim 5, characterized in that: The acquisition module specifically includes: An image acquisition unit, used to acquire a plurality of images of tea leaves to be tested at different angles; A spectrum acquisition unit, used to collect spectrum data of the tea leaves to be tested within a preset wavelength range; The electronic nose collection unit is used to collect electrical signal data with aroma characteristics.

7. The tea detection device according to claim 5, characterized in that: The extraction module specifically includes: A determination unit, used to determine the connection relationship of the edge pixels of the tea leaves based on the images of the tea leaves to be detected at different angles and a preset convolutional neural network architecture learning model; A calculation and generation unit is used to calculate and generate the ratio of the perimeter and area of ​​the outline of the tea leaves to be detected based on the connection relationship of the edge pixels of the tea leaves. ; An extraction unit, used for extracting texture complexity features from images of tea leaves to be detected at different angles; The first generating unit is used to generate the entropy value of the gray-level symbiosis matrix of the tea to be detected based on the gray-level symbiosis matrix algorithm. ; The first calculation unit is used to calculate the average slope of the preset wavelength range based on the spectral data of the tea leaves to be detected in the preset wavelength range. ; The second calculation unit is used to calculate the relative content of the targeted aroma component in the total aroma components based on the electrical signal data with aroma characteristics. .

8. The tea detection device according to claim 5, characterized in that: The identification module specifically includes: Judgment unit, used to judge the fusion feature value Is it greater than the characteristic value threshold? The second generation unit is used to fuse the eigenvalues If the value is greater than the characteristic value threshold, a rejection control instruction is generated.

9. A computer storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 4 is implemented.

Images