Longjing tea fresh leaf fine sorting method based on machine vision feedback

Abstract

The present application belongs to the technical field of tea processing equipment, and particularly relates to a Longjing tea fresh leaf fine sorting method based on machine vision feedback, comprising a fresh leaf conveying mechanism, a machine vision detection mechanism, a quality analysis mechanism, a sorting execution mechanism and a feedback adjustment mechanism; the method comprises: conveying fresh leaves to be sorted, collecting fresh leaf image information, processing the image and determining the quality grade of the fresh leaves, and controlling the fresh leaves to enter the corresponding collection area according to the quality grade; obtaining the actual quality information after sorting, and adjusting the image collection parameters, quality analysis parameters or sorting execution parameters according to the deviation between the actual quality information and the quality grade, to realize feedback optimization. The present application can reduce the influence of fresh leaf stacking and shielding on visual detection, improve the consistency of quality judgment, and improve the ability of the equipment to adapt to different fresh leaf batches and changes in operating environment.

Description

Technical Field

[0001] This invention belongs to the technical field of tea processing equipment, specifically referring to a method for fine sorting of fresh Longjing tea leaves based on machine vision feedback. Background Technology

[0002] Longjing tea, a traditional and renowned Chinese tea, relies heavily on the quality of its fresh leaves for subsequent processing and the final product. During processing, the fresh leaves are typically categorized based on factors such as tenderness, leaf integrity, color, and size to ensure the consistency of subsequent processing steps.

[0003] Currently, the sorting of fresh Longjing tea leaves is mainly done manually. Workers judge the quality grade of the leaves by observing their appearance and combining this with their experience. However, the existing manual sorting method has the following shortcomings: First, manual sorting is inefficient and cannot meet the needs of continuous processing during peak harvesting periods; second, manual judgment is easily affected by differences in worker experience, leading to inconsistent sorting results for the same batch of leaves; furthermore, some existing automated sorting equipment is designed for dry tea or granular materials, while Longjing tea leaves are characterized by their softness, significant size differences, and tendency to stick together or pile up during transport. If existing visual inspection methods are used directly, some leaf areas may be obscured, resulting in incomplete image acquisition; additionally, existing visual sorting equipment typically uses fixed detection parameters, requiring manual readjustment of equipment parameters when ambient lighting conditions or batch sizes change, affecting the stability of continuous operation. Summary of the Invention

[0004] To address the aforementioned challenges, this invention provides a method for fine sorting of fresh Longjing tea leaves based on machine vision feedback. By setting up a uniform material mechanism before the fresh leaves enter the visual detection area, the piled-up fresh leaves are dispersed and sorted, reducing mutual occlusion between leaves. Combined with machine vision detection, quality analysis, and feedback adjustment, the parameters in the continuous sorting process of Longjing tea leaves are optimized.

[0005] To achieve the above functions, the technical solution adopted by the present invention is as follows: a method for fine sorting of fresh Longjing tea leaves based on machine vision feedback, including a fresh leaf conveying mechanism, a machine vision detection mechanism, a quality analysis mechanism, a sorting execution mechanism, and a feedback adjustment mechanism. The fresh leaf conveying mechanism is used to convey the Longjing tea fresh leaves to be sorted, and to make the Longjing tea fresh leaves pass through the detection area and the sorting area in sequence; The machine vision inspection mechanism is set in the inspection area and is used to collect image information of fresh Longjing tea leaves in the conveying state; The quality analysis mechanism is connected to the machine vision inspection mechanism and is used to obtain the color features, morphological features, size features and surface condition features of fresh leaves based on the collected fresh leaf image information, generate quality evaluation results based on the above feature information, and determine the quality grade of fresh leaves according to the preset quality grade evaluation standard. The comprehensive evaluation parameters consist of fresh leaf color consistency index, leaf integrity index, size matching index, and surface defect ratio index. The quality analysis agency calculates the above indicators to obtain evaluation results for distinguishing different quality grades of fresh leaves. The sorting execution mechanism is connected to the quality analysis mechanism and is used to control the classification and output of Longjing tea fresh leaves of different quality grades according to the quality grade. The feedback adjustment mechanism is communicatively connected to the machine vision inspection mechanism, the quality analysis mechanism, and the sorting execution mechanism, respectively, to obtain the sampling inspection results of the sorted fresh leaves, compare the sampling inspection results with the quality grade output by the quality analysis mechanism to obtain the feedback deviation, and adjust the uniform material parameters, image acquisition parameters, quality evaluation parameters, or sorting execution parameters according to the feedback deviation to achieve parameter correction of the continuous sorting process.

[0006] As a preferred embodiment of the present invention, the fresh leaf conveying mechanism includes a conveyor belt and a material leveling mechanism disposed at the input end of the conveyor belt. The material leveling mechanism is used to disperse and organize the Longjing tea fresh leaves entering the conveyor belt, so that the fresh leaves entering the detection area have less mutual occlusion area, which facilitates the machine vision detection mechanism to obtain effective image information.

[0007] As a preferred embodiment of the present invention, the material equalization mechanism includes a vibrating feeding plate, a vibrating drive connected to the vibrating feeding plate, and a material limiting structure disposed at the outlet position of the vibrating feeding plate; the vibrating drive is used to drive the vibrating feeding plate to vibrate, so that the stacked Longjing tea leaves move and disperse; the material limiting structure is used to limit the thickness of the fresh leaves entering the conveyor belt, thereby reducing the probability of multiple fresh leaves being stacked at the same time and entering the detection area.

[0008] As a preferred embodiment of the present invention, the machine vision inspection mechanism includes an image acquisition device, an auxiliary light source, and a control unit; the image acquisition device is disposed at a corresponding position in the inspection area and is used to acquire image information of fresh Longjing tea leaves under transport conditions; the auxiliary light source is disposed around the inspection area and is used to provide a stable lighting environment to the inspection area; the control unit is connected to the image acquisition device and is used to control the image acquisition device to acquire fresh leaf images at a preset acquisition frequency.

[0009] As a preferred embodiment of the present invention, the quality analysis mechanism includes an image processing module and a quality recognition module; the image processing module is used to perform image enhancement, target region separation and feature extraction processing on the acquired fresh leaf images; the quality recognition module is used to establish quality evaluation results based on the extracted color features, morphological features, size features and surface condition features of the fresh leaves, and the quality recognition module compares the quality evaluation results with the pre-stored evaluation standards corresponding to different quality grades, and determines the quality grade corresponding to the fresh leaves based on the comparison results.

[0010] As a preferred embodiment of the present invention, when the quality identification module makes a comprehensive judgment: When all visual features correspond to the same quality level, that quality level is used as the final judgment result; when different visual features correspond to different quality levels, a comprehensive judgment is made based on the degree of conformity between each visual feature and the evaluation criteria.

[0011] As a preferred embodiment of the present invention, the sorting execution mechanism includes a pneumatic injection assembly and multiple collection areas; the pneumatic injection assembly includes multiple jet units spaced apart along the fresh leaf conveying direction, each jet unit corresponding to a collection area of ​​a different quality grade; the sorting execution mechanism determines the jet timing based on the fresh leaf quality grade and the position information of the fresh leaves during the conveying process, and controls the corresponding jet unit to output airflow so that the fresh leaves enter the collection area of ​​the corresponding quality grade.

[0012] As a preferred embodiment of the present invention, the feedback adjustment mechanism includes a result detection module and a parameter adjustment module; the result detection module is used to obtain the actual quality information of the sorted fresh leaves; The parameter adjustment module determines whether the deviation exceeds the allowable range based on the deviation between the actual quality information and the quality grade, according to a preset deviation judgment rule. When the deviation exceeds the allowable range, the parameter adjustment module adjusts the corresponding parameters according to the deviation type. The corresponding parameters include image acquisition parameters, quality evaluation parameters, and sorting execution parameters.

[0013] Specifically, when the deviation is caused by changes in image acquisition quality, adjust the image acquisition parameters; when the deviation is caused by inaccurate quality grade judgment, adjust the quality evaluation parameters; and when the deviation is caused by fresh leaves entering the wrong collection area, adjust the sorting execution parameters.

[0014] As a preferred technical solution of the present invention, when the proportion of fresh leaves whose actual test results are inconsistent with the quality grade output by the quality identification module 19 exceeds a preset threshold, it is determined that there is a deviation; the preset threshold is set according to production requirements.

[0015] Based on the above system, the present invention also provides a method for fine sorting of fresh Longjing tea leaves based on machine vision feedback, comprising the following steps: S1. The fresh Longjing tea leaves to be sorted are transported to the testing area and dispersed and sorted by the material equalization mechanism. S2. Collect image information of fresh Longjing tea leaves within the inspection area using a machine vision inspection mechanism; S3. Process the collected fresh leaf images to obtain the color features, morphological features, size features and surface condition features of the fresh leaves. S4. Establish quality evaluation parameters based on the color characteristics, morphological characteristics, size characteristics and surface condition characteristics, and determine the corresponding quality grade of Longjing tea fresh leaves based on the quality evaluation parameters. S5. Control the sorting mechanism to operate according to the quality grade, so that fresh leaves of different quality grades enter the corresponding collection area; S6. Obtain the sampling results of the sorted fresh leaves and compare the sampling results with the quality grade output by the quality analysis agency 17 to obtain the feedback deviation amount. S7. Adjust the corresponding operating parameters according to the feedback deviation.

[0016] Compared with the prior art, the present invention achieves the following beneficial effects by adopting the above structure: By employing a feeding method that combines a vibrating feeder with a limiting structure, the fresh Longjing tea leaves are distributed with low overlap before entering the inspection area. This effectively improves the problem of incomplete image acquisition caused by leaf occlusion, providing a reliable image foundation for subsequent visual inspection. By comprehensively evaluating four indicators—color consistency, leaf integrity, size matching, and surface defects—the system avoids interference from fluctuations in a single characteristic in quality judgment. Simultaneously, the feedback adjustment mechanism can differentiate and adjust image acquisition parameters, quality evaluation parameters, and sorting execution parameters based on the sampling results after sorting. This eliminates the need for manual adjustments when changing batches of fresh leaves or when ambient light conditions change, improving the stability of continuous production. The above-mentioned processes of homogenization, testing, evaluation, sorting, and feedback work together to form a closed-loop sorting process suitable for the characteristics of Longjing tea leaves. This reduces the degree of manual intervention while improving the consistency of sorting results for the same batch of fresh leaves. Attached Figure Description

[0017] Figure 1 This is a block diagram of the overall structure of the Longjing tea fresh leaf fine sorting system based on machine vision feedback of the present invention; Figure 2 This is a schematic diagram of the structure of the fresh leaf conveying mechanism and the uniform feeding mechanism of the present invention; Figure 3 This is a flowchart of the quality analysis mechanism of the present invention; Figure 4 This is a flowchart of the feedback adjustment mechanism of the present invention.

[0018] The components include: 1. Fresh leaf conveying mechanism; 2. Conveyor belt; 3. Material leveling mechanism; 4. Vibrating feeding plate; 5. Vibration drive component; 6. Material limiting structure; 7. Frame; 8. Drive assembly; 9. Machine vision inspection mechanism; 10. Image acquisition device; 11. Auxiliary light source; 12. Control unit; 13. Sorting execution mechanism; 14. Pneumatic spray assembly; 15. Collection area; 16. Position detection unit; 17. Quality analysis mechanism; 18. Image processing module; 19. Quality recognition module; 20. Feedback adjustment mechanism; 21. Result detection module; and 22. Parameter adjustment module. Detailed Implementation

[0019] The specific embodiments of the present invention will be described below with reference to the accompanying drawings. The described embodiments are only for illustrating the present invention and do not constitute a limitation on the scope of protection. Example 1

[0020] like Figure 1 As shown, this embodiment provides a method and system for fine sorting of fresh Longjing tea leaves based on machine vision feedback, including a fresh leaf conveying mechanism 1, a machine vision detection mechanism 9, a quality analysis mechanism 17, a sorting execution mechanism 13, and a feedback adjustment mechanism 20.

[0021] The fresh leaf conveying mechanism 1 is used to carry and transport the Longjing tea fresh leaves to be sorted, so that the fresh leaves pass through the detection area and the sorting area in sequence. The machine vision detection mechanism 9 is set in the detection area on the conveying path to collect image information of the fresh leaves when they pass through the detection area. The quality analysis mechanism 17 is connected to the machine vision detection mechanism 9 to receive the collected fresh leaf images and analyze them according to the visual information such as the color, shape, size and surface condition of the fresh leaves in the images to determine the corresponding quality grade of the fresh leaves.

[0022] In this embodiment, the quality analysis agency 17 does not make a judgment based on a single visual parameter, but combines the actual appearance characteristics of Longjing tea leaves to comprehensively evaluate multiple visual feature information in order to reduce the judgment bias caused by changes in a single feature.

[0023] The sorting execution mechanism 13 is located behind the detection area and is used to classify and output fresh leaves of different quality grades according to the quality grade information output by the quality analysis mechanism 17. The feedback adjustment mechanism 20 is connected to the machine vision detection mechanism 9, the quality analysis mechanism 17 and the sorting execution mechanism 13 respectively, and is used to obtain the actual sorting results and adjust the corresponding operating parameters according to the difference between the actual sorting results and the quality analysis results.

[0024] In this embodiment, the fresh leaf conveying mechanism 1 realizes continuous conveying of fresh leaves, the material leveling mechanism 3 appropriately disperses and sorts the fresh leaves before conveying to reduce the accumulation of fresh leaves and mutual occlusion, the machine vision detection mechanism 9 acquires fresh leaf image information, the quality analysis mechanism 17 completes quality judgment, the sorting execution mechanism 13 completes classification output, and the feedback adjustment mechanism 20 corrects the parameters during operation.

[0025] Through the cooperation of the above-mentioned mechanisms, the present invention forms a continuous sorting process consisting of fresh leaf input, dispersion and sorting, visual inspection, quality judgment, automatic sorting, and feedback adjustment. Example 2

[0026] like Figure 2 As shown, the fresh leaf conveying mechanism 1 includes a conveyor belt 2, a material leveling mechanism 3, a frame 7, and a drive assembly 8.

[0027] The conveyor belt 2 is installed on the frame 7 and driven by the drive assembly 8 to transport the fresh Longjing tea leaves to be sorted.

[0028] In actual production, the fresh leaves of Longjing tea after picking are usually in a random pile state. If they directly enter the visual inspection area, some fresh leaves are prone to sticking together or stacking up and down, causing some leaf areas to be blocked during image acquisition. Therefore, in this embodiment, a material leveling mechanism 3 is set at the input end of the conveyor belt 2.

[0029] The material equalization mechanism 3 includes a vibrating feed plate 4, a vibrating drive component 5, and a material limiting structure 6. The vibrating feed plate 4 is inclined, with its inlet end used to receive fresh leaves to be sorted and its outlet end corresponding to the conveyor belt 2. The vibrating drive component 5 is installed below the vibrating feed plate 4 and is used to drive the vibrating feed plate 4 to generate periodic vibration. When the fresh leaves enter the vibrating feed plate 4, under the action of vibration force, the fresh leaves move relative to each other, causing some of the fresh leaves that are stuck together to separate, thereby reducing the degree of fresh leaf accumulation. The material limiting structure 6 is set at the outlet position of the vibrating feed plate 4.

[0030] In this embodiment, a gap is formed between the limiting structure 6 and the vibrating feeding plate 4 for fresh leaves to pass through. When the thickness of the fresh leaves piled up exceeds the set range, some fresh leaves are restricted by the limiting structure 6 and readjusted under the action of vibration. Through the cooperation of the vibrating feeding plate 4 and the limiting structure 6, the distribution of fresh leaves entering the conveyor belt 2 and passing through the detection area is more dispersed, reducing the occlusion between leaves. Example 3

[0031] like Figure 2 As shown, the machine vision inspection mechanism 9 includes an image acquisition device 10, an auxiliary light source 11, and a control unit 12.

[0032] The image acquisition device 10 is positioned above the detection area, with its lens facing the conveyor belt 2, and is used to acquire images of fresh leaves passing through the detection area.

[0033] In this embodiment, the image acquisition device 10 can be an industrial camera. The image acquisition device 10 is installed according to the width of the conveyor belt 2 and the size of the fresh leaves, so that the acquisition range covers the fresh leaf conveying area within the detection area. The auxiliary light source 11 is set around the image acquisition device 10 to supplement the detection area. Since the color of Longjing tea fresh leaves is easily affected by ambient light, the auxiliary light source 11 provides relatively stable lighting conditions to reduce the impact of external light changes on the image acquisition results.

[0034] The auxiliary light source 11 can be a ring light source, a strip light source, or a surface light source. The control unit 12 is connected to the image acquisition device 10 and the auxiliary light source 11 respectively. During the operation of the conveyor belt 2, the control unit 12 controls the image acquisition device 10 to continuously acquire data according to the conveying speed, so that the fresh leaves passing through the detection area can be completely recorded.

[0035] In this embodiment, the control unit 12 can adjust the image acquisition frequency according to the conveying speed to ensure that fresh leaf image information that meets the quality analysis requirements can be obtained under different conveying speed conditions. Example 4

[0036] like Figure 3 As shown, the quality analysis unit 17 includes an image processing module 18 and a quality recognition module 19.

[0037] The image processing module 18 receives fresh leaf images collected by the machine vision inspection mechanism and processes the images.

[0038] Specifically, it includes: First, the acquired images are denoised to reduce the impact of noise generated during image acquisition. Secondly, the image is enhanced to increase the difference between the fresh leaf area and the background area; Subsequently, the fresh leaf regions in the image are identified to obtain the corresponding fresh leaf target regions; Finally, corresponding appearance features are extracted based on the target area of ​​the fresh leaves.

[0039] In this embodiment, the extracted features include: (1) Color characteristics; used to reflect the color state of fresh leaves, such as color depth, color uniformity and color distribution.

[0040] Specifically, color characteristics can be obtained by analyzing color information within the fresh leaf area, such as obtaining the average color value, the range of color variation, and the color distribution, as a basis for quality evaluation.

[0041] This embodiment does not directly judge the quality of fresh leaves based on a single color parameter, but combines other visual characteristics for comprehensive evaluation, so as to reduce the impact of color differences caused by different harvesting times or environmental factors on the judgment results.

[0042] (2) Morphological characteristics; used to reflect the external outline of fresh leaves, such as the integrity of the leaf outline, the continuity of the edge, and the changes in leaf shape.

[0043] Specifically, leaf shape characteristics are obtained by analyzing the edge information of the target area of ​​fresh leaves.

[0044] Among them, morphological features can be used to determine whether there are obvious damaged areas in fresh leaves or whether the outline is missing due to shading.

[0045] (3) Size characteristics; used to reflect the size information of fresh leaves. Specifically, parameters such as leaf length, width and projected area are obtained through the target area segmentation results and used as the basis for quality evaluation.

[0046] (4) Surface condition features; used to reflect abnormal conditions on the surface of fresh leaves, such as damaged areas, abnormal color areas, etc. The quality identification module 19 generates quality evaluation results based on the above visual feature information.

[0047] Specifically, the quality identification module 19 has pre-stored standard feature information corresponding to different quality grades. The standard feature information is established by collecting fresh tea leaf samples of different quality grades and combining them with the results of manual grading. The quality identification module 19 compares the obtained visual feature information of the fresh leaves with the pre-stored evaluation standards and determines the corresponding quality grade of the fresh leaves based on the comparison results.

[0048] The process of forming the comprehensive evaluation parameters is as follows: each indicator is standardized and calculated according to a preset weight to obtain the corresponding evaluation value; the corresponding quality level is determined according to the interval of the evaluation value; when multiple visual features correspond to the same quality level, the quality level is used as the final judgment result; when different visual features correspond to different quality levels, a comprehensive judgment is made based on the degree of conformity between each visual feature and the evaluation standard.

[0049] For example, when the color and size characteristics of fresh leaves meet the requirements of a higher quality grade, but the surface condition detection finds obvious damaged areas, the quality identification module 19 lowers the corresponding grade based on the comprehensive evaluation results.

[0050] The above method enables a comprehensive judgment based on multiple appearance characteristics of fresh leaves, improving the consistency of judgment in the sorting process of different batches of fresh leaves. Example 5

[0051] like Figure 2As shown, the sorting actuator 13 includes a pneumatic injection assembly 14 and multiple collection areas 15.

[0052] The pneumatic injection assembly 14 includes multiple injection units arranged at intervals along the fresh leaf conveying direction. Each injection unit is connected to a control valve to generate a low-pressure pulse airflow according to a control signal.

[0053] In this embodiment, the sorting execution mechanism 13 further includes a position detection unit 16, which is located between the detection area and the sorting area. It is used to detect the movement position of the fresh leaves on the conveyor belt 2 and generate a corresponding position signal. The position detection unit 16 can be a photoelectric sensor, a visual positioning module, or other detection device that can detect the movement position of the fresh leaves. When the quality identification module 19 determines a certain quality grade of a fresh leaf, the sorting execution mechanism 13 determines the time when the fresh leaf arrives at the corresponding jet unit based on the quality grade information recorded when the fresh leaf is identified in the detection area and the position information detected by the position detection unit 16. When the fresh leaf moves to the position of the corresponding jet unit, the control valve opens, causing the jet unit to output airflow. After being acted upon by the airflow, the fresh leaf changes its direction of movement and enters the collection area 15 of the corresponding quality grade. Since the jet unit and the fresh leaf adopt a non-contact sorting method, the damage to the fresh leaf caused by mechanical contact can be reduced.

[0054] In this embodiment, the sorting execution mechanism 13 determines the jet delay time based on the running speed of the conveyor belt 2 and the distance between the fresh leaf detection position and the jet unit.

[0055] When the feedback adjustment mechanism 20 detects an increase in the proportion of fresh leaves entering the erroneous collection area 15, it can adjust the jet delay time or jet duration to match the jet action with the actual movement position of the fresh leaves. Example 6

[0056] like Figure 4 As shown, the feedback adjustment mechanism 20 includes a result detection module 21 and a parameter adjustment module 22.

[0057] During system operation, the sorted fresh leaves enter the corresponding collection area 15, and the result detection module 21 is used to obtain the actual quality information of the sorted fresh leaves.

[0058] In this embodiment, the actual quality information can be obtained through manual sampling. In a more automated implementation, the actual quality information can also be obtained through a detection device set in the collection area 15.

[0059] The parameter adjustment module 22 compares the actual quality information with the quality level output by the quality recognition module 19, and obtains the feedback deviation amount based on the difference between the two. The feedback deviation amount is used to represent the degree of difference between the machine vision judgment result and the actual detection result.

[0060] Specifically, staff members extract fresh leaf samples from different collection areas 15 at a set frequency and input them into the result detection module 21 according to the actual quality. The result detection module 21 compares the actual detection results with the quality grade output by the quality identification module 19 and calculates the feedback deviation. When there is a significant difference between the actual detection results and the output results of the quality identification module 19, the parameter adjustment module 22 determines the adjustment object according to the deviation type.

[0061] Specifically, it includes: (1) When multiple quality levels show recognition deviation, it may be due to changes in the image acquisition environment. In this case, adjust the image acquisition parameters. (2) When image acquisition is normal but there is a deviation in quality grade judgment, adjust the quality evaluation parameters; (3) When the quality judgment is correct but the fresh leaves enter the wrong collection area, adjust the sorting execution parameters.

[0062] The image acquisition parameters include exposure time, gain parameters, and auxiliary light source brightness parameters. Quality evaluation parameters include color evaluation weight, size evaluation threshold, and surface defect judgment threshold; The sorting execution parameters include jet delay time and jet duration.

[0063] In addition, in some embodiments, the parameter adjustment module 22 can also adaptively adjust the operating parameters of the uniform feeding mechanism according to the distribution of fresh leaves in the detection area.

[0064] For example, when an increase in the occlusion of fresh leaves in the detection area is detected, the vibration frequency of the vibration drive 5 can be adjusted to improve the distribution of fresh leaves before they enter the detection area.

[0065] When the proportion of fresh leaves whose actual test results do not match the quality grade output by the quality identification module 19 exceeds a preset threshold, a deviation is determined.

[0066] The parameter adjustment module 22 adjusts the corresponding parameters according to the deviation and re-tests the sorting effect after adjustment.

[0067] Through the above feedback adjustment method, the system can correct relevant parameters according to the actual operating status, thereby improving the stability of the continuous sorting process. Example 7

[0068] This embodiment provides a method for fine sorting of fresh Longjing tea leaves based on machine vision feedback, including: S1. The fresh Longjing tea leaves to be sorted are fed into the fresh leaf conveying mechanism 1. After the fresh leaves enter the uniform material mechanism 3, they are dispersed and moved under the action of the vibrating feeding plate 4, and the degree of fresh leaf accumulation is reduced by the limiting material structure 6. S2. Collect image information of fresh Longjing tea leaves in the detection area through machine vision inspection mechanism 9; S3. The quality analysis unit 17 receives the collected images and processes the fresh leaf images through the image processing module 18 to obtain the color characteristics, morphological characteristics, size characteristics and surface condition characteristics of the fresh leaves. S4. The quality identification module 19 generates quality evaluation results based on the acquired visual feature information and determines the quality grade of fresh leaves according to the preset evaluation standards. S5. The sorting actuator 13 controls the corresponding jet unit to operate according to the quality grade, so that fresh leaves of different quality grades enter the corresponding collection area 15. S6. The feedback adjustment mechanism 20 obtains the sampling inspection results of the sorted fresh leaves and compares the sampling inspection results with the quality grade output by the quality identification module 19 to obtain the feedback deviation amount. S7, Parameter Adjustment Module 22 adjusts the image acquisition parameters, quality evaluation parameters, or sorting execution parameters according to the feedback deviation.

[0069] In practical use, staff first put the picked Longjing tea leaves into the fresh leaf conveying mechanism 1. After the fresh leaves enter the uniform feeding mechanism 3, they move under the action of the vibrating feeding plate 4, which disperses some of the piled-up fresh leaves and reduces the stacking of fresh leaves by the limiting structure 6.

[0070] Subsequently, conveyor belt 2 carries the fresh leaves through the inspection area, and machine vision inspection mechanism 9 collects images of the fresh leaves.

[0071] The quality analysis unit 17 processes the images to obtain visual characteristic information of fresh leaves and determines the quality grade according to the preset evaluation criteria.

[0072] When the fresh leaves move to the sorting area, the sorting actuator 13 controls the corresponding jet unit to operate according to the quality grade, so that the fresh leaves enter the corresponding collection area 15.

[0073] During continuous operation of the equipment, staff regularly conduct random checks on the fresh leaves within the collection area 15.

[0074] The feedback adjustment mechanism 20 adjusts the relevant operating parameters based on the difference between the sampling inspection results and the quality judgment results.

[0075] The present invention and its embodiments have been described above. This description is not restrictive, and the accompanying drawings are only one embodiment of the present invention; the actual structure is not limited thereto. In conclusion, if those skilled in the art are inspired by this description and design similar structures and embodiments without departing from the spirit of the invention, such designs should fall within the protection scope of the present invention.

Claims

1. A fine sorting system for fresh Longjing tea leaves based on machine vision feedback, characterized in that: It includes a fresh leaf conveying mechanism (1), a machine vision inspection mechanism (9), a quality analysis mechanism (17), a sorting execution mechanism (13), and a feedback adjustment mechanism (20); The fresh leaf conveying mechanism (1) is used to convey the Longjing tea fresh leaves to be sorted, and to make the Longjing tea fresh leaves pass through the detection area and the sorting area in sequence; The machine vision inspection mechanism (9) is set in the inspection area and is used to collect image information of fresh Longjing tea leaves under the transport state; The quality analysis mechanism (17) is connected to the machine vision inspection mechanism (9) for obtaining the color features, morphological features, size features and surface condition features of fresh leaves based on the collected fresh leaf image information, and generating quality evaluation results based on the above feature information, and matching the quality evaluation results with the preset quality grade evaluation standard to determine the quality grade of fresh leaves. The sorting execution mechanism (13) is communicatively connected to the quality analysis mechanism (17) and is used to control the classification and output of Longjing tea leaves of different quality grades according to the quality grade. The feedback adjustment mechanism (20) is connected to the machine vision inspection mechanism (9), the quality analysis mechanism (17) and the sorting execution mechanism (13) respectively. It is used to obtain the sampling inspection results of the fresh leaves after sorting, compare the sampling inspection results with the quality grade output by the quality analysis mechanism (17) to obtain the feedback deviation amount, and adjust the uniform material parameters, image acquisition parameters, quality evaluation parameters or sorting execution parameters according to the feedback deviation amount, so as to realize the parameter correction of the continuous sorting process.

2. The Longjing tea fresh leaf fine sorting system based on machine vision feedback according to claim 1, characterized in that: The fresh leaf conveying mechanism (1) includes a conveyor belt (2) and a uniform material mechanism (3) set at the input end of the conveyor belt (2). The uniform material mechanism (3) is used to disperse and arrange the Longjing tea fresh leaves entering the conveyor belt (2), reduce the degree of fresh leaf accumulation and mutual occlusion between leaves.

3. The Longjing tea fresh leaf fine sorting system based on machine vision feedback according to claim 2, characterized in that: The material leveling mechanism (3) includes a vibrating feeding plate (4), a vibrating drive component (5) connected to the vibrating feeding plate (4), and a material limiting structure (6) set at the outlet position of the vibrating feeding plate (4). The vibrating drive component (5) is used to drive the vibrating feeding plate (4) to vibrate, so that the stacked Longjing tea leaves move and disperse. The material limiting structure (6) is used to limit the thickness of the fresh leaves entering the conveyor belt (2), reducing the probability of multiple fresh leaves being stacked at the same time and entering the detection area.

4. The Longjing tea fresh leaf fine sorting system based on machine vision feedback according to claim 3, characterized in that: The machine vision inspection mechanism (9) includes an image acquisition device (10), an auxiliary light source (11), and a control unit (12). The image acquisition device (10) is located at the corresponding position of the inspection area and is used to acquire image information of fresh Longjing tea leaves under the transport state. The auxiliary light source (11) is located around the inspection area and is used to provide a stable lighting environment to the inspection area. The control unit (12) is connected to the image acquisition device (10) and is used to control the image acquisition device (10) to acquire fresh leaf images according to a preset acquisition frequency.

5. The Longjing tea fresh leaf fine sorting system based on machine vision feedback according to claim 1, characterized in that: The quality analysis unit (17) includes an image processing module (18) and a quality recognition module (19); the image processing module (18) is used to perform image enhancement, target region separation and feature extraction processing on the collected fresh leaf images; The quality identification module (19) is used to establish quality evaluation results based on the extracted color features, morphological features, size features and surface condition features of fresh leaves. The quality identification module (19) compares the quality evaluation results with the pre-stored evaluation standards corresponding to different quality grades, and determines the quality grade corresponding to the fresh leaves based on the comparison results.

6. The Longjing tea fresh leaf fine sorting system based on machine vision feedback according to claim 5, characterized in that: When the quality identification module (19) makes a comprehensive judgment: When all visual features correspond to the same quality level, that quality level is used as the final judgment result; when different visual features correspond to different quality levels, a comprehensive judgment is made based on the degree of conformity between each visual feature and the evaluation criteria.

7. The Longjing tea fresh leaf fine sorting system based on machine vision feedback according to claim 6, characterized in that: The sorting execution mechanism (13) includes a pneumatic jet assembly (14) and multiple collection areas (15); the pneumatic jet assembly (14) includes multiple jet units spaced apart along the fresh leaf conveying direction, each jet unit corresponding to a collection area (15) of a different quality grade; the sorting execution mechanism (13) determines the jet timing according to the fresh leaf quality grade and the position information of the fresh leaves during the conveying process, and controls the corresponding jet unit to output airflow so that the fresh leaves enter the collection area (15) of the corresponding quality grade.

8. The Longjing tea fresh leaf fine sorting system based on machine vision feedback according to claim 1, characterized in that: The feedback adjustment mechanism (20) includes a result detection module (21) and a parameter adjustment module (22); the result detection module (21) is used to obtain the actual quality information of the sorted fresh leaves; The parameter adjustment module (22) determines whether the deviation exceeds the allowable range according to the deviation between the actual quality information and the quality grade and the preset deviation judgment rule. When the deviation exceeds the allowable range, the parameter adjustment module (22) adjusts the corresponding parameters according to the deviation type. The corresponding parameters include image acquisition parameters, quality evaluation parameters, and sorting execution parameters.

9. The Longjing tea fresh leaf fine sorting system based on machine vision feedback according to claim 1, characterized in that: When the proportion of fresh leaves whose actual test results are inconsistent with the quality grade output by the quality identification module (19) exceeds the preset threshold, it is determined that there is a deviation; the preset threshold is set according to production requirements.

10. A method for fine sorting of fresh Longjing tea leaves based on machine vision feedback according to any one of claims 1 to 9, characterized in that: Includes the following steps: S1. The fresh Longjing tea leaves to be sorted are transported to the testing area and dispersed and sorted by the equalization mechanism (3); S2. Collect image information of fresh Longjing tea leaves in the detection area through the machine vision inspection mechanism (9); S3. Process the collected fresh leaf images to obtain the color features, morphological features, size features and surface condition features of the fresh leaves. S4. Establish quality evaluation parameters based on the color characteristics, shape characteristics, size characteristics and surface condition characteristics, and determine the corresponding quality grade of Longjing tea fresh leaves based on the quality evaluation parameters. S5. Control the sorting execution mechanism (13) to operate according to the quality grade, so that fresh leaves of different quality grades enter the corresponding collection area (15). S6. Obtain the sampling results of the sorted fresh leaves, and compare the sampling results with the quality grade output by the quality analysis agency (17) to obtain the feedback deviation amount. S7. Adjust the corresponding operating parameters according to the feedback deviation.

Images