Machine vision-based intelligent sorting system and method for machine-harvested tea fresh leaves

Abstract

The present application relates to the technical field of postharvest sorting of tea fresh leaves, and discloses a machine-vision-based intelligent sorting system and method for machine-harvested tea fresh leaves, which comprises a feeding and dispersing module, a primary sorting module, an intelligent sorting module and a database module. The method uses the feeding and dispersing module to spread the machine-harvested tea fresh leaves on the negative pressure conveying unit and scatter them to the primary sorting module by the discharging unit; the primary sorting module preliminarily classifies the machine-harvested tea fresh leaves and sends them to the intelligent sorting module respectively; the intelligent sorting module collects the image information of the tea fresh leaves at different levels, extracts the feature data of the tea fresh leaf images, compares and analyzes the feature data with the corresponding tea fresh leaf information in the database module, generates the grade determination instruction of the tea fresh leaves and transmits the sorting instruction, and the sorting mechanism executes the sorting decision to complete the classification. The present application combines the intelligent classification technology with the physical classification technology, realizes the efficient and accurate intelligent sorting while ensuring the minimum damage and loss of the tea fresh leaves.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of postharvest sorting of tea fresh leaves, in particular to a machine-vision-based intelligent sorting system and method for machine-picked tea fresh leaves. BACKGROUND

[0002] Tea is a beverage raw material made from the tender buds, leaves or stems of the tea plant, rich in various health-benefiting components such as tea polyphenols, caffeine, amino acids, and vitamins. By brewing tea water, it provides a refreshing, thirst-quenching and invigorating effect, and also plays an important social and etiquette role in different cultures.

[0003] Currently, tea picking is generally carried out by machines, especially in large-scale tea gardens, which significantly improves the picking efficiency and yield. Machine-picked tea leaves use advanced mechanical equipment to detach tea leaves from tea plants through vibration, shearing, and other methods. Although mechanical picking has obvious advantages in speed and cost, the quality and uniformity of tea leaves may decrease due to the inability to finely select tea bud leaves as in manual picking.

[0004] Although some sorting equipment has emerged in the current market to replace manual labor and achieve large-scale sorting of machine-picked tea fresh leaves, in actual application, such equipment has potential risks in the sorting process, which may damage the tea fresh leaves. More critically, there are often many residual leaves and fragments mixed in the screening results, affecting the overall quality. In addition, these devices have insufficient capacity in tea fresh leaf grading, often requiring operators to perform secondary grading on the sorted tea fresh leaves, thereby increasing labor and time costs.

[0005] Therefore, there is an urgent need to invent a sorting technology for machine-picked tea fresh leaves to solve the problem of existing tea leaf sorting equipment that easily damages tea fresh leaves during the sorting process, has poor sorting effect, and requires secondary sorting by manual labor, thereby increasing labor and time costs. SUMMARY

[0006] In view of this, the present application proposes a machine-vision-based intelligent sorting system and method for machine-picked tea fresh leaves, which overcomes the problem of low precision, large loss, and the need for secondary sorting by manual labor, thereby increasing labor and time costs.

[0007] The present application proposes a machine-vision-based intelligent sorting system for machine-picked tea fresh leaves, comprising:

[0008] The upper feeding discrete module includes a negative pressure conveying unit and a lower feeding unit. The negative pressure conveying unit adsorbs the tea fresh leaves in the upper feeding area to the conveying belt and moves to the lower feeding unit with the conveying belt. The lower feeding unit area has no negative pressure environment, and the tea fresh leaves fall freely to the next module due to their own gravity.

[0009] The primary sorting module is connected with the feeding discrete module, and includes a sorting unit, a shaking unit and a recovery unit; wherein the sorting unit is composed of four sorting windows, each sorting window is composed of screen holes with different sizes and different densities, buffer belts are arranged between the sorting windows to isolate them, and the sorting is performed according to the shape characteristics of the tea leaves; the tea leaves of each level after sorting are sequentially dropped into the next module through the recovery unit of each level;

[0010] The intelligent sorting module is connected with the recovery unit of the primary sorting module, adopts a multi-level channel cooperative sorting method, and includes an image acquisition unit, an image transmission unit, an image analysis unit, a control unit and an airflow blowing unit; wherein the image acquisition unit acquires image information of the tea leaves, and transmits the image information to the image analysis unit, the image analysis unit determines the types of the tea leaves and generates a determination instruction which is transmitted to the control unit, the control unit instructs the airflow blowing unit to operate, and the airflow blowing unit performs the operation to complete accurate grading of the tea leaves;

[0011] The database module includes a computer and a data transmission unit, a data triggering unit, a data storage unit and a data indexing unit, and is used for storing and calling image information of different forms of tea leaves of each level of each variety, and storing newly acquired image information of the tea leaves.

[0012] Further, the feeding discrete module includes:

[0013] The negative pressure conveying unit is composed of a conveying belt, a negative pressure cavity and an air compressor, wherein small holes with uniform spacing are arranged on the conveying belt, the diameter of the small holes is 1mm, and each small hole in the negative pressure cavity can adsorb the tea leaves;

[0014] The discharging unit is composed of an isolation cavity and a conveying belt, when the conveying belt passes through the isolation cavity, the negative pressure effect is removed, the tea leaves are no longer adsorbed, and the tea leaves are free falling to the next unit;

[0015] The feeding area of the entire feeding discrete module is larger than the discharging area, so as to ensure that the tea leaves are quickly and dispersedly sorted.

[0016] Further, the primary sorting module includes:

[0017] The sorting unit includes a vibrating screening device and sorting windows of tea leaves of each level, wherein the inclination angle of the vibrating screening device is between [25°, 40°], and the vibration frequency is between [25Hz, 35Hz];

[0018] The screen surface of the vibrating screening device is composed of four sorting windows with different screen holes arranged at different densities, and the areas of the sorting windows are different, wherein the areas gradually decrease from the first sorting area to the fourth sorting area;

[0019] The inclination angle, vibration frequency, and the mesh size, distribution area, and the area of the buffer zone of the sorting window of each level of tea leaves are adjusted according to the type and characteristics of the tea fresh leaves, and the reference characteristic indexes are the length, spread, weight, and rolling friction coefficient of the tea fresh leaves.

[0020] The recycling unit is connected to each sorting window and receives each level of tea fresh leaves after sorting, and uses the gravity of the tea fresh leaves to transmit them one by one to the next module through a smooth track.

[0021] Further, the intelligent sorting module comprises:

[0022] The image acquisition unit comprises a trigger device and a high-speed camera device, and when the tea fresh leaves pass through, the trigger device sends a signal to the high-speed camera device for image acquisition, and the frame rate of the high-speed camera device cannot be lower than 100 fps.

[0023] The image analysis unit comprises a computer and image preprocessing algorithms, feature extraction algorithms, pattern recognition algorithms, decision-making algorithms, and data transmission interfaces, and is associated with the database module; wherein the image analysis unit performs preprocessing such as noise reduction, enhancement, greying, and segmentation on the received images, extracts basic morphological features of the tea fresh leaves, including basic morphological features such as area, perimeter, major axis, and minor axis, and complex morphological features including grey value, rectangularity, circularity, compactness, and fine length; by calling the image features of tea fresh leaves of different grades of the variety in the database module, the collected tea fresh leaves are determined and a decision instruction is made.

[0024] The control unit comprises a computer, a data transmission channel, and a control switch, and is connected to the image analysis unit, receives the determination instruction of the image analysis unit, and transmits data information to the control switch.

[0025] The airflow blowing unit is connected to the control switch and comprises a blowing device and a gas valve switch, and blows out airflow of a certain intensity according to the instruction of the control unit.

[0026] Compared with the prior art, the machine vision-based machine-harvested tea leaf intelligent sorting system has the beneficial effects that: the tea leaves are evenly distributed and in a non-overlapping state through the feeding and dispersing module, providing ideal input conditions for subsequent sorting and effectively avoiding recognition errors caused by tea leaf stacking in traditional sorting processes. Meanwhile, the primary sorting module realizes multi-level grading according to the shape characteristics of the tea leaves, enabling different levels of tea leaves to enter the next link in an orderly manner and laying a foundation for fine sorting. Such a grading method not only optimizes the sorting process but also reduces the complexity and errors of preliminary screening. Secondly, the intelligent sorting module introduces machine vision technology, combined with the image acquisition, transmission and analysis unit, to realize efficient determination and accurate grading of tea leaves. The airflow blowing unit performs sorting operations according to the instructions of the control unit, so that each tea leaf can be accurately classified. This image analysis-based sorting method not only improves sorting accuracy but also adapts to the morphological differences of different varieties of tea leaves, with high flexibility and expandability. In addition, the addition of the database module enables the system to store a large amount of tea leaf image information, forming a comprehensive knowledge base for multiple categories and supporting dynamic updating and calling of new data.

[0027] The application also provides a machine vision-based machine-harvested tea leaf intelligent sorting method, comprising the following steps:

[0028] S1: According to the shape characteristics of machine-harvested tea leaves, including leaf length, leaf spread, weight and rolling friction coefficient, the tea leaves are preliminarily sorted into grade I, grade II, grade III, grade IV, grade V and others;

[0029] S2: Through a multi-channel acquisition mode, the image information of grade I-IV tea leaves is collected respectively, and the tea leaf feature data is extracted after image preprocessing, including basic morphological features (A) such as area (A1), perimeter (A2), major axis (A3) and minor axis (A4), and complex morphological features (B) such as gray value (B1), rectangularity (B2), circularity (B3), compactness (B4) and fine length (B5);

[0030] S3: Extract the important features of the tea leaves, calculate the correlation degree of each feature in the basic morphological features and complex morphological features of the tea leaves with the corresponding features in the database, and extract the features with the largest correlation degree respectively;

[0031] S4: According to the features extracted in S3, the tea leaves of each grade are determined and selected through the correlation algorithm. If the tea leaves meet the grade, no rejection instruction is given; if the tea leaves do not meet the grade, a rejection instruction is given. The sorted tea leaves are divided into selected grade I, selected grade II, selected grade III and selected grade IV;

[0032] S5: After the tea leaves of each grade are rejected, they are collected and subjected to secondary sorting.

[0033] Further, in the step S1, the tea fresh leaves are preliminarily classified into Grade I, Grade II, Grade III, Grade IV, Grade V and other standards;

[0034] Among them: the main component of Grade I tea fresh leaves is single bud; the main component of Grade II tea fresh leaves is one bud and one leaf; the main component of Grade III tea fresh leaves is one bud and two leaves; the main component of Grade IV tea fresh leaves is one bud and three leaves; and the main component of Grade V tea fresh leaves is one bud and four leaves or one bud and five leaves.

[0035] Further, in the step S3, the important feature extraction method of the tea fresh leaves is:

[0036] The basic morphological feature data (A1-A4) and the complex morphological feature data (B1-B5) of the tea fresh leaves are respectively calculated with the corresponding feature data of the tea fresh leaves of each grade of the variety in the database module to obtain the corresponding correlation coefficients kA1-kA4 and kB1-kB5, wherein 1, 2, 3, 4 and 5 represent each grade of tea leaves, i.e. Grade I, Grade II, Grade III, Grade IV and Grade V and other standards, kAi is the correlation coefficient of the basic morphological feature of the tea fresh leaves and the corresponding feature data of the sample in the database, kAi∈[0,1], i=1, 2, 3, 4; kBj is the correlation coefficient of the complex morphological feature of the tea fresh leaves and the corresponding feature data of the sample in the database, kBj∈[0,1], j=1, 2, 3, 4, 5; two feature data sets with the largest correlation are extracted respectively: Amax∈{A1,A2,A3,A4} and Bmax∈{B1,B2,B3,B4,B5}.

[0037] Further, in the step S4, the determination method of the grade of the tea fresh leaves is:

[0038] The determination coefficient K=kA*0.6+kB*0.4 is calculated, wherein kA is the correlation coefficient obtained by calculating Amax, and kB is the correlation coefficient obtained by calculating Bmax; when K>K0, it is determined that the tea fresh leaves meet the grade, and no rejection instruction is given; when K≤K0, it is determined that the tea fresh leaves do not meet the grade, and a rejection instruction is given, wherein K0∈[0.6,1], and the value is determined according to the type of tea, the correlation degree of the characteristics and the classification accuracy requirement.

[0039] Compared with the prior art, the beneficial effects of the machine vision-based intelligent sorting method for machine-harvested tea fresh leaves of the present application are that: in the preliminary sorting stage, the tea fresh leaves are classified based on shape feature parameters (such as leaf length, leaf spread, weight and rolling friction coefficient), and the tea leaves are preliminarily classified into grades I-V and other categories. This process simplifies the complexity of subsequent sorting, provides a clear basis for further fine classification, and reduces the burden of data processing. Secondly, in the image information acquisition and analysis stage, through multi-channel acquisition mode and image preprocessing technology, basic morphological features including area, perimeter, major and minor axis, and complex morphological features such as gray value, rectangularity and circularity can be extracted from tea fresh leaves. These features are matched with the standard data in the database for correlation, and the most recognizable features are selected for accurate judgment. This design based on multi-dimensional feature extraction and correlation analysis greatly improves the scientificity and reliability of the sorting results, and can adapt to the differentiated characteristics of various tea leaves, improving the intelligent level of sorting. Finally, through the correlation algorithm and the execution of the sorting instruction, the tea fresh leaves can be accurately selected and rejected, and the rejected part can be sorted again to reduce resource waste. This double sorting mechanism not only guarantees the quality of high-grade tea fresh leaves, but also effectively utilizes low-grade and marginal tea leaves, maximizing the sorting value. BRIEF DESCRIPTION OF DRAWINGS

[0040] Various other advantages and benefits will become apparent to those of ordinary skill in the art upon reading the following detailed description of the preferred embodiments. The drawings are for purposes of illustration only and are not intended to limit the present application thereto. Moreover, like reference numerals in the drawings designate similar parts throughout the several views. In the drawings:

[0041] Figure 1 A functional block diagram of a machine vision-based intelligent sorting system for machine-harvested tea fresh leaves is provided for embodiments of the present application.

[0042] Figure 2 A flowchart of a machine vision-based intelligent sorting method for machine-harvested tea fresh leaves is provided for embodiments of the present application.

[0043] Figure 3 A structural schematic diagram of a machine vision-based intelligent sorting system for machine-harvested tea fresh leaves is provided for embodiments of the present application.

[0044] Figure 4 A structural schematic diagram of a primary sorting module is provided for embodiments of the present application.

[0045] 100, an upper material discrete module; 200, a primary sorting module; 220, a vibrating screening device; 230, a recovery unit; 231, a first recovery window; 232, a second recovery window; 233, a third recovery window; 234, a fourth recovery window; 235, a fifth recovery window; 300, an intelligent sorting module; 310, a first sorting end; 320, a second sorting end; 330, a third sorting end; 340, a fourth sorting end; 350, a fifth sorting end. DETAILED DESCRIPTION

[0046] Exemplary embodiments of the present disclosure will be described in detail with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the drawings, it is understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided so that the present disclosure can be more thoroughly understood, and the scope of the present disclosure can be accurately conveyed to those skilled in the art. It should be noted that the embodiments in the present disclosure and the features in the embodiments can be combined with each other without conflict. The present disclosure will be described in detail below with reference to the accompanying drawings and in conjunction with embodiments.

[0047] Tea leaves are a beverage raw material made from the tender buds, leaves, or stems of the tea plant, rich in various health-benefiting components such as tea polyphenols, caffeine, amino acids, and vitamins. By brewing tea water, it provides a refreshing, thirst-quenching, and invigorating effect, while also playing an important social and etiquette role in different cultures.

[0048] Currently, tea picking is generally carried out by machines, especially in large-scale tea plantations, which significantly improves picking efficiency and yield. Machine-picked tea leaves use advanced mechanical equipment to remove tea leaves from tea plants through vibration, shearing, and other methods. Although mechanical picking has obvious advantages in speed and cost, the quality and uniformity of tea leaves may decrease due to the inability to finely select tea leaf buds as in manual picking.

[0049] Although some sorting equipment has emerged in the current market, aiming to replace manual work to achieve large-scale sorting of machine-picked tea leaves, in actual application, such equipment has potential risks in the sorting process, which may damage tea leaves. More critically, there are often many residual leaves and fragments mixed in the screening results, affecting the overall quality. In addition, these devices are still insufficient in their ability to grade tea leaves, often requiring operators to perform secondary grading on the sorted tea leaves, thereby increasing labor and time costs.

[0050] In view of this, the present application proposes a machine vision-based machine-harvested tea fresh leaf intelligent sorting system and method, aiming to solve the problem that the current technology of tea leaf sorting equipment is prone to damage tea fresh leaves during sorting, and the sorting effect is poor, and secondary sorting is needed through manual operation, thereby increasing labor cost and time cost.

[0051] As Figure 1 shown, in some embodiments of the present application, the present embodiment provides a machine vision-based machine-harvested tea fresh leaf intelligent sorting system, which comprises a feeding discrete module 100, a primary sorting module 200, an intelligent sorting module 300 and a database module.

[0052] Specifically, the feeding discrete module 100 comprises a negative pressure conveying unit and a discharging unit; wherein the negative pressure conveying unit disperses and adsorbs the tea fresh leaves in the feeding area to the conveying belt and moves to the discharging unit with the conveying belt, and the discharging unit area has no negative pressure environment, and the tea fresh leaves fall freely to the next module due to their own gravity; the primary sorting module 200 is connected with the feeding discrete module 100, comprising a sorting unit, a shaking unit and a recovery unit 230; wherein the sorting unit is graded according to the shape characteristics of tea fresh leaves; the sorted tea fresh leaves of each level fall into the next module one by one through the recovery unit 230 of each level; the intelligent sorting module 300 is connected with the recovery unit 230 of the primary sorting module 200, adopts a multi-level channel cooperative sorting method, and comprises an image acquisition unit, an image transmission unit, an image analysis unit, a control unit and an airflow blowing unit; wherein the image acquisition unit acquires image information of tea fresh leaves and transmits it to the image analysis unit, determines the type of tea fresh leaves and generates a determination instruction to the control unit, the control unit instructs the airflow blowing unit to operate, and the airflow blowing unit executes the operation to complete the accurate grading of tea fresh leaves; the database module comprises a computer and a data transmission unit, a data triggering unit, a data storage unit and a data indexing unit, which is used to store and call the image information of different forms of tea fresh leaves of each level of each variety, and store the image information of newly collected tea fresh leaves.

[0053] It can be understood that the tea fresh leaves are dispersed and uniformly adsorbed onto the conveying belt in a non-overlapping flat state by the negative pressure conveying unit in the feeding and dispersing module 100. This design effectively avoids the problem of leaf stacking affecting the subsequent sorting accuracy, provides standardized initial conditions for the sorting module, and ensures the running stability of the entire system. Secondly, in the primary sorting module, the tea leaves are preliminarily classified according to their shape characteristics (such as length, width, etc.) by using a sorting unit composed of 4-level sorting windows, and different grades of tea leaves are respectively dropped into the next module by the recovery unit 230. This module uses a step-by-step sorting method to preliminarily classify tea fresh leaves according to basic morphological characteristics, providing basic data for subsequent more refined sorting. This grading mechanism combined with the coordinated operation of the sorting unit and the conveying unit enables the system to efficiently process large quantities of tea fresh leaves while reducing the complexity of the primary sorting link. Finally, the intelligent sorting module 300 and the database module together realize high-precision classification of tea fresh leaves. The intelligent sorting module 300 obtains image information of tea fresh leaves through the image acquisition unit, extracts multi-dimensional features using the image analysis unit, and compares and analyzes the standard morphological data of each grade of tea fresh leaves stored in the database module. Finally, the control unit generates operation instructions to drive the airflow blowing unit to complete the grading operation. This intelligent sorting method combining machine vision and database not only enables accurate determination of tea fresh leaf types and grades, but also dynamically updates and calls newly collected image data, enhancing the system's adaptability to multiple varieties of tea leaves.

[0054] Specifically, the feeding and dispersing module 100 includes a negative pressure conveying unit composed of a conveying belt, a negative pressure cavity and an air compressor, wherein the conveying belt is arranged with uniformly spaced small holes, the small holes have a diameter of 1 mm, and each small hole in the negative pressure cavity can adsorb tea fresh leaves; the discharging unit is composed of an isolation cavity and a conveying belt, when the conveying belt passes through the isolation cavity, the negative pressure effect is removed, the tea fresh leaves are no longer adsorbed, and fall freely to the next unit; the feeding area of the entire feeding and dispersing module 100 is larger than the discharging area to ensure the rapid dispersion of the tea fresh leaves.

[0055] It can be understood that the negative pressure conveying unit utilizes the synergy of the conveying belt, the negative pressure cavity and the air compressor to perform negative pressure adsorption on the tea fresh leaves through the uniformly arranged small holes on the conveying belt. Each small hole has a diameter of 1 mm, which can generate a negative pressure effect suitable for adsorbing tea fresh leaves, ensuring that the tea fresh leaves can be firmly adsorbed on the surface of the conveying belt. This adsorption mechanism not only avoids the problem of tea fresh leaves slipping or stacking during conveying, but also realizes stable conveying and preliminary arrangement of tea leaves, providing an orderly basis for subsequent sorting. Secondly, the discharging unit realizes the precise release of tea fresh leaves through the structural design of the isolation cavity. When the conveying belt passes through the isolation cavity, the negative pressure effect is removed, and the tea fresh leaves are separated from the conveying belt and fall into the next unit in the form of free fall. This process fully utilizes mechanical design and air pressure principles, precisely controls the timing of adsorption and release, and ensures that tea fresh leaves can be dispersed according to the specified path and state. This not only ensures the uniformity of tea fresh leaf discharging, but also avoids the accumulation or damage that may be caused by improper release. Finally, the overall module realizes high-efficiency dispersion effect through the optimized design of the feeding area and the discharging area. The design of the feeding area being larger than the discharging area increases the distribution space of tea fresh leaves during negative pressure adsorption, thereby reducing the probability of tea fresh leaf accumulation. The gradually shrinking discharging area further concentrates and uniformly releases the tea fresh leaves, ensuring the unity of dispersion efficiency and quality.

[0056] Specifically, the primary sorting module 200 includes a sorting unit including a vibrating sieve device 220 and sorting windows of various levels of tea leaves, wherein the inclination angle of the vibrating sieve device 220 is between [25°, 40°], and the vibration frequency is between [25Hz, 35Hz]; the sieve surface of the vibrating sieve device is composed of four sorting windows with different sieve hole sizes and different density arrangements; the areas of the sorting windows are different, and gradually decrease from the first sorting area to the fourth sorting area; the inclination angle, the vibration frequency, and the sieve hole size, distribution area, and buffer belt area of the sorting windows of various levels of tea leaves are adjusted according to the types and characteristics of tea fresh leaves, and the reference characteristic indicators are leaf length, leaf spread, weight, and rolling friction coefficient of tea fresh leaves; the recovery unit 230 is connected with the sorting windows of various levels, respectively receives the sorted tea fresh leaves of various levels, and uses the self-gravity of the tea fresh leaves to transmit them one by one to the next module through a smooth track.

[0057] Preferably, the sorting windows of various levels are specifically the first recovery window 231, the second recovery window 232, the third recovery window 233, the fourth recovery window 234, and the fifth recovery window 235.

[0058] It can be understood that by setting the screen surface inclination angle of the vibrating screening device 220 between 25° to 40°, combined with the variation of the vibration frequency in the range of 25 Hz to 35 Hz, the tea fresh leaves are dispersed onto the screen surface using mechanical vibration and gravity. By controlling the vibration frequency and inclination angle, the tea fresh leaves move in an orderly manner on the screen surface and slide towards the corresponding sorting window. This vibrating screening method not only effectively reduces the possibility of tea fresh leaf accumulation, but also flexibly adjusts the parameters according to the leaf characteristics, ensuring the efficiency and accuracy of the sorting process. Secondly, the design of the multi-stage sorting window utilizes the principle of gradual change of screen hole density and area, achieving the grading and screening of tea fresh leaves. The screen surface consists of four-stage sorting windows with different screen hole densities, with the screen hole size gradually decreasing from the first to the fourth stage, and the sorting window area also gradually decreasing, allowing tea fresh leaves to pass through different windows according to their size, shape, weight, and other characteristics. The design parameters of the screen holes and windows can be dynamically adjusted according to the characteristics of the tea fresh leaves (such as leaf length, leaf spread, weight, rolling friction coefficient, etc.), enhancing the adaptability of the system to different types of tea fresh leaves. This gradual screening mechanism can quickly separate tea fresh leaves of different grades, while ensuring the accuracy and efficiency of the screening results. Finally, the recovery unit 230 achieves efficient transportation of sorted tea fresh leaves through gravity transmission and track guidance. The tea fresh leaves from each sorting window are received by the recovery unit 230, which uses the gravity of the tea fresh leaves to naturally slide down a smooth track and be transported to the next module one by one. This design simplifies the transportation process of tea fresh leaves, avoids damage to tea leaves caused by unnecessary mechanical operations, and ensures the separation status of tea fresh leaves of different grades, ensuring the working efficiency and accuracy of subsequent sorting modules.

[0059] Specifically, the intelligent sorting module 300 includes: an image acquisition unit including a trigger device and a high-speed camera device, which emits a signal to the high-speed camera device for image acquisition when tea fresh leaves pass through; the frame rate of the high-speed camera device cannot be lower than 100 fps; an image analysis unit including a computer and image preprocessing algorithms, feature extraction algorithms, pattern recognition algorithms, decision-making algorithms, and data transmission interfaces, which is associated with a database module; wherein the image analysis unit preprocesses the received images through noise reduction, enhancement, grayscale, and segmentation, extracts the basic morphological features of tea fresh leaves, including basic morphological features such as area, perimeter, major axis, minor axis, etc., and complex morphological features including gray value, rectangularity, circularity, compactness, and length, etc.; determines the collected tea fresh leaves by calling the image features of tea fresh leaves of different grades of the same variety in the database module, and makes a decision instruction; a control unit including a computer, a data transmission channel, and a control switch, which is connected to the image analysis unit, receives the determination instruction from the image analysis unit, and transmits data information to the control switch; an air blowing unit connected to the control switch, including a blowing device and an air valve switch, which blows air of a certain intensity according to the instruction of the control unit.

[0060] It can be understood that the image acquisition unit is composed of a triggering device and a high-speed camera. When the tea fresh leaves pass through, the triggering device sends a signal to activate the high-speed camera for image acquisition. The frame rate of the high-speed camera is set to no less than 100 fps, which can ensure clear capture of fast-moving tea fresh leaves. This combination of triggering and high-speed shooting ensures the timeliness and stability of image acquisition, providing high-quality raw data for subsequent feature extraction and analysis. Secondly, the image analysis unit realizes the automatic grading of tea fresh leaves through multi-level image processing and intelligent algorithms. This unit uses preprocessing algorithms (such as noise reduction, enhancement, grayscale, segmentation) to optimize the collected images, and extracts basic morphological features (such as area, perimeter, major axis, minor axis, etc.) and complex morphological features (such as gray value, rectangularity, circularity, compactness, fine length, etc.) of tea fresh leaves through feature extraction algorithms. Subsequently, the image analysis unit calls the characteristic information of tea fresh leaves of the same variety stored in the database module, combines pattern recognition algorithms and decision-making algorithms to determine the tea fresh leaves, and generates decision instructions. This design based on big data and intelligent algorithms not only improves the accuracy of sorting, but also adapts to the differences in characteristics of different tea varieties. Finally, the control unit and the airflow blowing unit realize the sorting of tea fresh leaves through intelligent instruction execution and airflow driving. After receiving the determination instructions transmitted by the image analysis unit, the control unit transmits signals to the air valve switch of the airflow blowing unit. Through the cooperation of the air blowing device and the air valve switch, the airflow blowing unit blows out airflow of appropriate intensity according to the sorting needs of tea fresh leaves of different grades, completing the accurate grading of tea fresh leaves. Through the combination of this mechanical and algorithmic design, the system can real-time and efficiently separate tea fresh leaves of different grades, providing high-quality sorting results for subsequent processing, while realizing the full automation and intelligentization of the sorting process.

[0061] Preferably, the airflow blowing unit is provided with a plurality of sorting ends for secondary selection and grading of tea fresh leaves from each sorting window, wherein the plurality of sorting ends at least include a first sorting end 310, a second sorting end 320, a third sorting end 330, a fourth sorting end 340, and a fifth sorting end 350.

[0062] Compared with the prior art, the machine vision-based machine-harvested tea leaf intelligent sorting system of the application provides ideal input conditions for subsequent sorting by uniformly distributing and presenting the tea leaves in a non-overlapping state through the feeding dispersion module 100, effectively avoiding recognition errors caused by tea leaf stacking in the traditional sorting process. At the same time, the primary sorting module 200 realizes multi-level grading according to the shape characteristics of the tea leaves, enabling different levels of tea leaves to enter the next link in an orderly manner, laying the foundation for fine sorting. Such a grading method not only optimizes the sorting process, but also reduces the complexity and errors of preliminary screening. Secondly, the intelligent sorting module 300 introduces machine vision technology, combined with image acquisition, transmission and analysis units, to realize efficient determination and accurate grading of tea leaves. The airflow blowing unit performs sorting operations according to the instructions of the control unit, so that each tea leaf can be accurately classified. This image analysis-based sorting method not only improves the sorting accuracy, but also adapts to the morphological differences of different varieties of tea leaves, with high flexibility and expandability. In addition, the addition of the database module enables the system to store a large amount of tea leaf image information, forming a comprehensive knowledge base for multiple categories and supporting dynamic updating and calling of new data.

[0063] Based on the above, Figure 2 The application also provides a machine vision-based machine-harvested tea leaf intelligent sorting method, which comprises the following steps:

[0064] S1: According to the shape characteristics of machine-harvested tea leaves, including the length, spread, weight and rolling friction coefficient of tea leaves, the tea leaves are preliminarily sorted into grade I, grade II, grade III, grade IV, grade V and others.

[0065] Specifically, in step S1, the standards for preliminarily sorting tea leaves into grade I, grade II, grade III, grade IV, grade V and others; wherein: the main component of grade I tea leaves is single bud; the main component of grade II tea leaves is one bud and one leaf; the main component of grade III tea leaves is one bud and two leaves; the main component of grade IV tea leaves is one bud and three leaves; the main component of grade V tea leaves is one bud and four leaves and one bud and five leaves.

[0066] It can be understood that through the analysis of physical parameters such as leaf length, leaf spread, weight and rolling friction coefficient, the classification of tea leaves is realized. Specifically, these shape features are used as the basis to divide the development state of tea fresh leaves from bud to leaf into different grades. By quantitatively evaluating these features, the preliminary sorting of tea fresh leaves can be effectively realized, which divides them into grades I to V and other categories according to the number and development of buds and leaves. This method can provide accurate and reliable standards in the preliminary sorting stage, thereby providing data support for subsequent fine sorting. In the sorting standards, the division of grades I to V is based on clear criteria, representing different development degrees of tea fresh leaves. Grade I tea fresh leaves are mainly composed of single buds, which belong to the best quality category of tea leaves. Grades II to V tea fresh leaves represent single buds combined with different numbers of leaves, and the quality of these grades of tea leaves usually decreases with the increase of leaves. The more leaves, the longer the subsequent processing time required. Through this division based on the number of leaves, high-quality tea fresh leaves can be quickly and effectively screened and separated from lower-grade tea leaves, laying the foundation for the precision of post-sorting processing. Finally, through the combination of quantitative physical characteristics and sorting standards, this sorting method can adapt to different types of tea fresh leaves and differences in picking. By corresponding the physical characteristics of tea fresh leaves to the tea leaf grade standards, the sorting process is not only scientific and reasonable, but also flexible according to the specific characteristics of the picked tea leaves.

[0067] S2: Through the multi-channel acquisition mode, the image information of grades I-IV tea fresh leaves is collected respectively, and after image preprocessing, the tea fresh leaf feature data is extracted, including basic morphological features (A) such as area (A1), perimeter (A2), major axis (A3), minor axis (A4), etc., and complex morphological features (B) including gray value (B1), rectangularity (B2), circularity (B3), compactness (B4), fine length (B5), etc.

[0068] S3: Extract important features of tea fresh leaves, calculate the correlation degree of each feature in the basic morphological features and complex morphological features of tea fresh leaves with the corresponding features in the database, and extract the features with the largest correlation degree respectively.

[0069] Specifically, in step S3, the important feature extraction method of tea fresh leaves is: the basic morphological feature data (A1-A4) and the complex morphological feature data (B1-B5) of tea fresh leaves are respectively calculated with the corresponding feature data of each level of tea fresh leaves of the variety in the database module to obtain the corresponding correlation coefficients kA1-kA4, kB1-kB5, wherein 1, 2, 3, 4, 5 represent each level of tea, i.e. I, II, III, IV, V and others, kAi is the correlation coefficient of the basic morphological feature of tea fresh leaves and the corresponding feature data of the sample in the database, kAi∈[0,1], i=1, 2, 3, 4; kBj is the correlation coefficient of the complex morphological feature of tea fresh leaves and the corresponding feature data of the sample in the database, kBj∈[0,1], j=1, 2, 3, 4, 5; two feature data sets with the largest correlation are extracted respectively: Amax∈{A1, A2, A3, A4} and Bmax∈{B1, B2, B3, B4, B5}.

[0070] It can be understood that in step S3, the basic morphological features (such as area, perimeter, major axis, minor axis, etc.) and complex morphological features (such as gray value, rectangularity, circularity, compactness, fine length, etc.) of tea fresh leaves are first extracted, which are used for comparison with the corresponding data of tea fresh leaf samples in the database. By calculating the correlation coefficient (kA1-kB5) of each feature and the feature data of tea fresh leaves of the corresponding variety and level in the database, the system can quantify the degree of association between different features and tea fresh leaf varieties and levels, thereby providing a scientific basis for subsequent sorting. Secondly, the correlation coefficient calculation quantifies the degree of association between features, ensuring that the system can dynamically adjust according to the matching degree of the specific morphological features of tea fresh leaves and the samples in the database. The correlation coefficients kAi and kBj between features are between 0 and 1, representing the similarity of different features and sample data. In this way, the system can identify the features most closely related to the classification of tea fresh leaves from the basic morphological features (A series) and complex morphological features (B series), ensuring the accuracy and efficiency of feature selection. Finally, the process of extracting the maximum correlation features enables the system to accurately grade and judge based on the morphological feature data (Amax and Bmax) of tea fresh leaves. Amax and Bmax represent the maximum correlation features corresponding to the variety and level of tea fresh leaves in the basic morphological features and complex morphological features, respectively.

[0071] S4: According to the features extracted in S3, the correlation algorithm is used to determine and select each level of fresh tea leaves. If the tea fresh leaves meet the level, no rejection instruction is given; if the tea fresh leaves do not meet the level, a rejection instruction is given. The sorted tea leaves are divided into selected I, II, III, and IV grades.

[0072] Specifically, in step S4, the determination method of the tea leaf grade is as follows: a determination coefficient K is calculated, where kA is the correlation coefficient calculated from Amax, and kB is the correlation coefficient calculated from Bmax; when K > K0, it is determined that the tea leaf meets the grade, and no rejection instruction is given; when K ≤ K0, it is determined that the tea leaf does not meet the grade, and a rejection instruction is given, where K0 ∈ [0.6, 1], and the value is determined according to the tea type, the correlation degree of the characteristics, and the accuracy of the classification requirement.

[0073] It can be understood that in step S4, the grade determination of the tea leaf is achieved by calculating the determination coefficient K, which is obtained by weighted combination of the correlation coefficient of the basic morphological characteristics (kA) and the correlation coefficient of the complex morphological characteristics (kB). Specifically, kA and kB represent the correlation degrees of Amax and Bmax characteristics extracted in S3, respectively, which reflect the matching degree of the tea leaf with the samples in the database. By assigning 60% weight to kA and 40% weight to kB, the determination coefficient K obtained comprehensively reflects the performance of the tea leaf in morphological characteristics, thereby achieving accurate evaluation of the tea leaf grade. Secondly, the setting of the determination coefficient K ensures that the system can compare the characteristic values of the tea leaf with the standards in the database, and determine whether the tea leaf meets a specific grade by the set threshold K0. When K is greater than the threshold K0, it is determined that the tea leaf meets the grade, and no rejection is needed; when K is less than or equal to K0, it is considered that the tea leaf does not meet the grade, and the system generates a rejection instruction. This threshold K0 can be adjusted within the range of [0.6, 1], and is determined comprehensively according to the tea type, the correlation degree of the characteristics, and the accuracy of the classification requirement, thereby ensuring the accuracy and adaptability of the sorting. Finally, based on the sorting mechanism of the determination coefficient, the system can effectively sort the tea leaf into different grades. Through the determination coefficient K, the tea leaf will be accurately classified into grades such as selected grade I, grade II, grade III, and grade IV, and the tea leaf meeting the grade standard will be retained, while the tea leaf not meeting the standard will be rejected.

[0074] S5: After the tea leaves rejected in each grade are collected, secondary sorting is performed.

[0075] It can be understood that the machine vision-based intelligent sorting method for machine-harvested tea fresh leaves in the embodiment of the present application can classify the tea fresh leaves based on shape feature parameters (such as leaf length, leaf spread, weight, and rolling friction coefficient) in the preliminary sorting stage, and preliminarily divide the tea leaves into grades I-V and other categories. This process simplifies the complexity of subsequent sorting, provides a clear basis for further fine grading, and reduces the burden of data processing. Secondly, in the image information acquisition and analysis stage, through the multi-channel acquisition mode and image preprocessing technology, basic morphological features such as area, perimeter, major and minor axis, and complex morphological features such as gray value, rectangularity, and circularity can be extracted from the tea fresh leaves. These features are matched with the standard data in the database for correlation, and the most recognizable features are selected for accurate judgment. This design based on multi-dimensional feature extraction and correlation analysis greatly improves the scientificity and reliability of the sorting result, and can adapt to the differentiated characteristics of various tea leaves, improving the intelligent level of sorting. Finally, through the correlation algorithm and the execution of the sorting instruction, the tea fresh leaves can be accurately selected and rejected, and the rejected part can be sorted again to reduce resource waste. This double sorting mechanism not only guarantees the quality of high-grade tea fresh leaves, but also effectively utilizes low-grade and marginal tea leaves, maximizing the sorting value.

[0076] Those skilled in the art will appreciate that embodiments of the application can be provided as methods, systems or computer program products. Accordingly, the application can be embodied in the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the application can be embodied in the form of a computer program product on one or more computer-usable storage media (including, but not limited to, disk memory, CD-ROM, optical memory, etc.) having computer usable program code embodied therein.

[0077] The application is described with reference to flowcharts and / or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams can be implemented by computer program instructions, and the combination of flows and / or blocks in the flowcharts and / or block diagrams. These computer program instructions can be provided to a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing apparatus to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing apparatus produce a means for implementing the functions specified in the flowcharts and / or block diagrams. Figure 1 one or more flows and / or blocks Figure 1 an apparatus for performing the functions specified in one or more flows or blocks.

[0078] These computer program instructions can also be stored in a computer readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer readable memory produce an article of manufacture including instructions which implement the flow Figure 1 of the flow or flows and / or blocks Figure 1 of the block or blocks specified in the flow.

[0079] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the flow Figure 1 of the flow or flows and / or blocks Figure 1 of the block or blocks specified in the flow.

[0080] Finally, it should be noted that the above-mentioned embodiments are merely used to illustrate the technical solutions of the present application, rather than limit the present application. Although the present application has been described in detail with reference to the above-mentioned embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or equivalent replaced without departing from the spirit and scope of the present application, and any modification or equivalent replacement should be covered in the protection scope of the claims of the present application.

Claims

1. A machine vision-based intelligent sorting system for machine-harvested fresh tea leaves, characterized in that, include: The feeding and dispersing module includes a negative pressure conveying unit and a discharging unit; wherein, the negative pressure conveying unit disperses and adsorbs the fresh tea leaves in the feeding area onto the conveyor belt and moves with the conveyor belt to the discharging unit, the discharging unit area has no negative pressure environment, and the fresh tea leaves fall freely to the next module due to their own gravity. The primary sorting module, connected to the feeding and dispersing module, includes a sorting unit, a shaking unit, and a recycling unit. The sorting unit consists of four levels of sorting windows, each with sieve holes of different sizes and densities. Buffer zones are set between the sorting windows for isolation, and the tea leaves are graded sequentially according to their shape characteristics. After sorting, the tea leaves at each level pass through the recycling units at each level and fall into the next module one by one. The intelligent sorting module, connected to the recycling unit of the primary sorting module, employs a multi-level channel collaborative sorting method, including an image acquisition unit, an image transmission unit, an image analysis unit, a control unit, and an airflow blowing unit. The image acquisition unit acquires image information of the fresh tea leaves and transmits it to the image analysis unit, which determines the type of tea leaves and generates a determination command that is transmitted to each level of control unit. The control unit then instructs the airflow blowing unit to perform the operation, thereby completing the precise grading of the fresh tea leaves. The database module includes a computer and data transmission unit, a data triggering unit, a data storage unit, and a data indexing unit. It is used to store and retrieve image information of different forms of fresh tea leaves of various varieties and grades, as well as to store newly collected image information of fresh tea leaves. The discrete feeding module includes: The negative pressure conveying unit consists of a conveyor belt, a negative pressure chamber, and an air compressor. The conveyor belt has evenly spaced small holes with a diameter of 1 mm. Each small hole in the negative pressure chamber can adsorb fresh tea leaves. The feeding unit consists of an isolation chamber and a conveyor belt. When the conveyor belt passes through the isolation chamber, the negative pressure is released, and the fresh tea leaves are no longer adsorbed and fall freely to the next unit. The area of ​​the feeding zone of the entire feeding discrete module is larger than the area of ​​the unloading zone to ensure that the fresh tea leaves to be sorted are quickly dispersed. The sorting unit includes a vibrating screening device and sorting windows for each grade of fresh tea leaves, wherein the tilt angle of the vibrating screening device is between [25°, 40°] and the vibration frequency is between [25Hz, 35Hz]. The vibrating screening device has a screen surface composed of four sorting windows with different screen holes arranged at different densities. The area of ​​each sorting window is different, and the area gradually decreases from the first-level sorting area to the fourth-level sorting area. The tilt angle, vibration frequency, and the size, distribution area, and buffer zone area of ​​the sieve holes in the sorting windows for each level of fresh tea leaves are adjusted according to the type and characteristics of the fresh tea leaves. The reference characteristic indicators are the leaf length, leaf spread, weight, and rolling friction coefficient of the fresh tea leaves. The recycling unit is connected to the sorting windows at each level, and receives the sorted tea leaves at each level. Using the weight of the tea leaves themselves, they are transported one by one to the next module via a smooth track. The intelligent sorting module includes: The image acquisition unit includes a triggering device and a high-speed camera. When fresh tea leaves pass by, the triggering device sends a signal to the high-speed camera to acquire the image. The frame rate of the high-speed camera must not be lower than 100fps. The image analysis unit includes a computer and image preprocessing algorithms, feature extraction algorithms, pattern recognition algorithms, decision-making algorithms, and a data transmission interface. This image analysis unit is associated with the database module. After preprocessing the received image by noise reduction, enhancement, grayscale conversion, and segmentation, the image analysis unit extracts the basic morphological features of fresh tea leaves. The basic morphological features of fresh tea leaves include area, perimeter, major axis, and minor axis, as well as complex morphological features, including grayscale value, rectangularity, roundness, compactness, and thinness. The image analysis unit uses the image features of different grades of tea leaves from the database module to determine the quality of the collected tea leaves and make decision instructions. The control unit includes a computer, a data transmission channel, and a control switch. It is connected to the image analysis unit and, after receiving the judgment instructions from the image analysis unit, transmits the data information to the control switch. The airflow blowing unit is connected to a control switch and includes an air blowing device and a valve switch. It blows out an airflow of a certain intensity according to the instructions of the control unit.

2. A machine vision-based intelligent sorting method for machine-harvested fresh tea leaves, applicable to the machine vision-based intelligent sorting system for machine-harvested fresh tea leaves as described in claim 1, characterized in that, include: S1: Based on the shape characteristics of machine-harvested tea leaves, including the number of leaf buds, leaf length, leaf spread, weight, and rolling friction coefficient, the tea leaves are initially sorted into Grade I, Grade II, Grade III, Grade IV, Grade V, and others; S2: Through multi-channel acquisition mode, image information of tea leaves of grades I-IV is acquired respectively. After image preprocessing, the feature data of tea leaves are extracted, including basic morphological features (A), such as area (A1), perimeter (A2), major axis (A3), minor axis (A4), and complex morphological features (B). Among them, complex morphological features (B) include gray value (B1), rectangularity (B2), roundness (B3), compactness (B4), and thinness (B5). S3: Extract the important features of fresh tea leaves, calculate the correlation between each feature in the basic and complex morphological features of fresh tea leaves and the corresponding features in the database, and extract the features with the highest correlation. S4: Based on the features extracted in S3, the tea leaves of each grade are identified and selected through the association algorithm. When the tea leaves are determined to meet a certain grade, the tea leaf sorting mechanism of the corresponding grade will be controlled to classify the tea leaves of that grade. The sorted tea leaves are divided into Selected Grade I, Selected Grade II, Selected Grade III, and Selected Grade IV.

3. The intelligent sorting method for machine-harvested fresh tea leaves based on machine vision according to claim 2, characterized in that, In step S3, the method for extracting the important features of the fresh tea leaves is as follows: The basic morphological feature data (A1~A4) and complex morphological feature data (B1~B5) of fresh tea leaves were compared with the corresponding feature data of each grade of fresh tea leaves of the same variety in the database module. The corresponding correlation coefficients kA1~kA4 and kB1~kB5 were calculated, where 1, 2, 3, 4, and 5 represent the grades of fresh tea leaves, namely Grade I, Grade II, Grade III, Grade IV, Grade V, and others. kAi is the correlation coefficient between the basic morphological features of fresh tea leaves and the corresponding feature data of the samples in the database, kAi∈[0,1], i=1, 2, 3, 4; kBj is the correlation coefficient between the complex morphological features of fresh tea leaves and the corresponding feature data of the samples in the database, kBj∈[0,1], j=1, 2, 3, 4, 5. The two feature datasets with the highest correlation were extracted: Amax∈{A1,A2,A3,A4} and Bmax∈{B1,B2,B3,B4,B5}.

4. The intelligent sorting method for machine-harvested fresh tea leaves based on machine vision according to claim 2, characterized in that, In step S4, the method for determining the grade of the fresh tea leaves is as follows: The determination coefficient K is calculated as K = kA * 0.6 + kB * 0.4, where kA is the correlation coefficient calculated by Amax and kB is the correlation coefficient calculated by Bmax. When K > K0, the tea leaves are determined to meet the grade and no rejection instruction is given. When K ≤ K0, the tea leaves are determined to not meet the grade and a rejection instruction is given. K0 ∈ [0.6, 1], and the value is determined by comprehensively considering the tea leaf type, characteristic correlation, and classification accuracy requirements.

Images