Intelligent photoelectric impurity removal device for tobacco lamina
By using an optical cross structure design with two pure black backgrounds at the top and bottom and an intelligent algorithm, the imaging quality problem caused by background contamination was solved, and efficient identification and removal of tobacco debris was achieved.
Patent Information
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- QILIN REDRYING FACTORY YUNNAN TOBACCO REDRYING
- Filing Date
- 2025-12-19
- Publication Date
- 2026-07-23
AI Technical Summary
Existing photoelectric impurity removal equipment suffers from poor imaging quality due to background plate contamination during the detection of tobacco materials, affecting the accuracy and stability of tobacco leaf and impurity identification.
It adopts an optical cross structure design with two pure black backgrounds at the top and bottom. It obtains a pure black background through the upper and lower imaging modules and parallel light source. Combined with RGB color recognition and AI debris detection algorithm, it identifies and removes debris.
It improved the quality of tobacco leaf image acquisition, reduced background interference, enhanced the accuracy and stability of debris identification, and increased the debris removal rate by 10%.
Smart Images
Figure CN2025143797_23072026_PF_FP_ABST
Abstract
Description
An intelligent sheet smoke photoelectric impurity removal device Technical Field
[0001] This invention relates to the field of tobacco impurity removal, and in particular to an intelligent sheet tobacco photoelectric impurity removal device based on a pure black background on both the upper and lower sides. Background Technology
[0002] Obtaining a pure black background has always been a technical challenge for existing photoelectric impurity removal equipment. Traditional photoelectric impurity removal equipment uses one of the three primary colors (R / G / B) as a background plate during the detection of tobacco materials. The target material quickly passes through the background plate, and a high-speed camera performs a line scan to acquire an image that includes both the tobacco leaves and the background plate. Image recognition technology is then used to distinguish between impurities and tobacco leaves, and the impurities are removed. However, in actual production operation, dust or other substances adhere to the background plate. After optical magnification, the image quality is greatly affected, and image processing to determine whether something is impurity or tobacco leaves is frequently interfered with, affecting the stability of the algorithm, resulting in a high false recognition rate and instability, thus impacting the impurity removal effect. Summary of the Invention
[0003] To address the technical problem in existing technologies where poor image quality during tobacco leaf scanning interferes with subsequent identification of tobacco leaves and impurities, this invention proposes an intelligent tobacco leaf photoelectric impurity removal device based on a dual-path pure black background. This invention abandons the traditional background plate design and innovatively adopts an optical cross-structure design, with high-speed scanning cameras positioned above and below the imaging background. This allows the device to simultaneously detect tobacco leaf materials during high-speed transport and acquire a pure black background, significantly reducing background interference during imaging, improving the quality of tobacco leaf image acquisition, and further enhancing the accuracy of impurity identification in tobacco leaves.
[0004] The specific plan is as follows:
[0005] An intelligent smoke and light removal device based on a dual-path pure black background includes:
[0006] Material transport module: includes a conveyor belt for transporting tobacco leaf materials;
[0007] Optical imaging modules: an upper imaging module that obtains tobacco leaf images by scanning the upper surface of the tobacco leaves with a camera, and a lower imaging module that obtains tobacco leaf images by scanning the lower surface of the tobacco leaves with a camera; the upper imaging module includes: a linear or area array camera that vertically captures images of the tobacco leaves below, a parallel light source that illuminates the upper imaging points of the tobacco leaf image from above for light compensation, and a slotted black background box for acquiring the black background of the upper tobacco leaf image; the lower imaging module includes: a linear or area array camera that vertically captures images of the tobacco leaves from the bottom, a parallel light source that illuminates the lower imaging points of the tobacco leaf image from below for light compensation, and a slotted black background box for acquiring the black background of the lower tobacco leaf image;
[0008] Image recognition module: includes: an image acquisition submodule for acquiring tobacco leaf images captured by the upper imaging module and the lower imaging module, and an image recognition submodule for identifying debris information in the tobacco leaf images; the image recognition submodule includes identifying debris information based on RGB color recognition algorithm and AI debris detection algorithm;
[0009] The rejection mechanism includes: a debris positioning submodule, a debris rejection submechanism, and a debris collection box; the debris positioning submodule locates the debris based on the debris information from the image recognition submodule and calculates the time it takes for the debris to reach the rejection mechanism; the debris rejection submechanism controls a solenoid valve to eject air and reject the debris when it reaches the rejection position after being thrown in a parabolic trajectory; the debris collection box is placed below the rejection position to receive the debris rejected by the solenoid valve.
[0010] Tobacco leaf collection mechanism: After removing impurities, the tobacco leaves fall into the tobacco leaf collection mechanism in a parabolic trajectory.
[0011] Preferably, the material transport module includes: a first conveyor belt for receiving tobacco leaf material and a second conveyor belt for spreading and accelerating the material; the end point of the first conveyor belt is located above the starting point of the second conveyor belt, and the tobacco leaf material is thrown in a parabolic trajectory and falls onto the starting point of the second conveyor belt.
[0012] Preferably, the material transport module further includes a fan structure for attaching tobacco leaf material to the surface of the conveyor belt; the fan mechanism includes: a first fan mechanism that blows air onto the upper layer of tobacco leaf material on the second conveyor belt to make the material evenly adhere to the conveyor belt; and a second fan mechanism that blows air in a direction perpendicular to the parabola of the tobacco leaf at the connection between the first and second conveyor belts to change the speed of the tobacco leaf.
[0013] Preferably, the optical imaging module includes: at least one set of parallel light sources: the width of one set of parallel light sources is set according to the width of the tobacco leaves on the conveyor belt, that is, the width of the parabola after the tobacco leaves are thrown out; if two sets of parallel light sources are selected, the two sets of parallel light sources are focused on the imaging point of the tobacco leaves to perform light-concentrating compensation on the tobacco leaves.
[0014] Preferably, the optical imaging module further includes a third fan structure that allows the tobacco leaves thrown from the material transport module to undergo parabolic motion at a certain speed, thereby changing the parabolic angle of the tobacco leaves as they fall.
[0015] Preferably, the optical imaging module further includes a glass assembly for isolating the actuator of the photoelectric cleaning device and the tobacco leaf, wherein the slotted black background box is embedded in the glass assembly, dividing the glass assembly into two pieces.
[0016] Preferably, the slotted black background box includes: a U-shaped slotted structure, a beveled triangular structure, and a U-shaped bottom roller structure; the slotted black background box has an opening at the top for light to enter, and a debris outlet at the bottom for dropping debris or a rotation control mechanism for removing debris by rotating the roller.
[0017] Preferably, the U-shaped groove structure further includes a light-blocking rod disposed at the focal point of the U-shaped arc.
[0018] Preferably, in the lower imaging module, the reflector is placed vertically at the bottom of the lower imaging point, and the linear or area array camera is placed at an angle, adjusting the tilt angle to capture the tobacco leaf image in the reflector; the angle between the camera and the reflector is adjustable.
[0019] Preferably, the image recognition submodule includes: an RGB color recognition algorithm unit that compares the colors of tobacco leaves and debris in an image based on RGB colors, calculates the probability of debris in the image, and identifies the debris using an RGB color recognition algorithm; an AI debris detection algorithm unit that trains an AI debris detection algorithm based on the debris image and detects whether debris exists in the tobacco leaf image based on the trained AI debris detection algorithm; and a debris information unit that locates the possible locations of debris based on the detection results of the RGB color recognition algorithm unit and the AI debris detection algorithm unit, and generates debris information.
[0020] Beneficial effects
[0021] This invention proposes an intelligent tobacco leaf photoelectric impurity removal device based on a dual-path pure black background. It designs an optical imaging module with dual-path imaging, improving image acquisition quality. This module not only acquires tobacco leaf images through both paths but also designs a focused parallel light source to improve brightness and uniformity while reducing light source attenuation per unit time. Thirdly, by designing a U-shaped groove black background plate structure, noise interference is reduced, further significantly improving the quality of the tobacco leaf images. Simultaneously, the high-quality tobacco leaf images also improve the performance and effectiveness of the RGB color recognition algorithm and the AI impurity detection algorithm, particularly providing a high-quality training set for the AI impurity detection algorithm, thus improving the accuracy of impurity detection. In summary, this invention proposes a highly efficient intelligent tobacco leaf photoelectric impurity removal device based on a dual-path pure black background. By obtaining high-quality tobacco leaf images and improving the recognition algorithm, it comprehensively improves the effect of impurity identification in tobacco leaves. Compared with traditional impurity removal equipment, this device improves the impurity removal rate by 10%.
[0022] The specific technical effects of this optical imaging module are as follows:
[0023] 1. Enhanced Contrast: This invention combines a black background with a parallel light source, resulting in higher contrast between the background and the tobacco leaves and debris. This makes it easier for the algorithm to distinguish the target object from its surrounding environment. High contrast highlights the outline and features of objects, which is highly beneficial for edge detection and feature extraction steps in the tobacco leaf removal process.
[0024] 2. Reduced Complexity: This invention constructs a clean black background through the design of the black background structure. The U-shaped trough background structure, including tilting and airflow elements, minimizes interfering factors in the scene. For example, cluttered background details might be mistakenly identified as part of the target, or cause unnecessary computational burden. By simplifying the background, the processing becomes more efficient, and resources can be focused on the area of actual interest.
[0025] 3. Improve signal-to-noise ratio: Ideal imaging conditions should have a high signal-to-noise ratio (i.e., the desired information) to noise (irrelevant or harmful information). By using specific designs to create near-ideal dark-field conditions, the influence of incoherent light can be effectively suppressed, thereby improving the quality of the final acquired data.
[0026] 4. Facilitates thresholding: Many image processing techniques rely on setting an appropriate threshold for binarization—converting an image into a black-and-white format. When the background is very dark, choosing an appropriate threshold becomes relatively easy, allowing for more accurate separation of foreground objects.
[0027] 5. Facilitates the application of deep learning algorithms in artificial intelligence. Real-world applications often present challenges under extreme lighting conditions, such as strong reflections or extremely dark areas. This invention optimizes the optical path design in the optical imaging module to obtain a uniform and controllable lighting environment, helping the camera sensor better handle changes in ambient light and avoiding overexposure or underexposure. A black background plays a crucial role in this regard due to its ability to absorb light.
[0028] 6. Adaptable to different lighting conditions: Even under unpredictable lighting conditions, the slotted black background box of the optical imaging module can always ensure a relatively stable dark background as a reference, which helps to maintain the consistency of system performance. Attached Figure Description
[0029] Figure 1 is a structural diagram of an intelligent sheet smoke photoelectric impurity removal device based on a pure black background on both the upper and lower sides according to the present invention.
[0030] Figure 2 is a schematic diagram of an optical imaging module based on a pure black background on both the top and bottom sides in an embodiment.
[0031] Figure 3 is a rendering of a U-shaped groove black background box structure design in the embodiment.
[0032] Figure 4 is a rendering of a design method for a beveled triangular black background box structure in one embodiment.
[0033] Figure 5 is a rendering of a U-shaped bottom roller-type black background box structure design in one embodiment.
[0034] Figure 6 shows an example of traditional background imaging effect in the embodiment.
[0035] Figure 7 shows the imaging effect obtained by scanning based on the optical imaging module of the present invention in the embodiment.
[0036] Explanation of reference numerals in the attached drawings: 1-Conveyor belt; 111-First conveyor belt; 112-Second conveyor belt; 2-Fan structure; 21 Upper airflow; 22-Second fan mechanism; 3-Parallel light source; 31-Lower parallel light source; 32-Upper parallel light source; 4-Glass assembly; 41-Upper glass assembly; 42-Lower glass assembly; 5-Trough-type black background box; 51-Upper trough-type black background box; 52-Lower trough-type black background box; 511-U-shaped trough-type Structure, 5111-Light blocking rod, 512-Hydraulic triangular structure, 5121-Air inlet, 5122-Dust and waste outlet, 513-U-shaped bottom roller structure; 5131-Rotating light blocking rod; 514-Light slit; 6-Linear or area array camera; 61-Upper line array or area array camera; 62-Lower line array or area array camera; 7-Reflector; 8-Light blocking plate; 9-Solenoid valve; 10-Miscellaneous waste collection box; 11-Tobacco leaf collection mechanism. Detailed Implementation
[0037] The specific embodiments of this invention will be further described below with reference to the accompanying drawings.
[0038] As shown in Figure 1, an intelligent smoke and light removal device based on a two-way pure black background includes:
[0039] Material transport module: includes: conveyor belt 1 for transporting tobacco leaf materials;
[0040] As shown in Figure 2, the optical imaging module comprises: an upper imaging module that scans the upper surface of the tobacco material to obtain an upper-path tobacco image, and a lower imaging module that scans the lower surface of the tobacco material to obtain a lower-path tobacco image; the upper imaging module includes: an upper parallel light source 32 that performs light compensation on the imaging point of the tobacco leaf, i.e., the position of the tobacco leaf to capture the tobacco image; an upper line array or area array camera 61 placed on the upper end of the tobacco material to capture the tobacco material vertically downwards; and a lower imaging module placed on the upper tobacco image... Below the imaging point, an upper channel slotted black background box 51 is used to capture the black background of the upper channel tobacco leaf image by the upper channel array or area array camera 61; the lower channel imaging module includes: a lower channel array or area array camera 62 that vertically captures the lower channel tobacco leaf image from the bottom; a lower channel parallel light source 31 that performs light compensation on the lower channel imaging point of the tobacco leaf, i.e., the position of the tobacco leaf in the tobacco leaf image; and a lower channel slotted black background box 52 placed above the imaging point of the tobacco leaf image so that the lower channel array or area array camera 62 can scan the image and capture the black background of the lower channel tobacco leaf image.
[0041] Image recognition module: includes: an image acquisition submodule that acquires the upper and lower tobacco leaf images from the upper and lower imaging modules to form a tobacco leaf image; and an image recognition submodule that identifies impurities in the tobacco leaf image and generates impurity information; the image recognition submodule includes impurity identification based on RGB color recognition algorithm and AI impurity detection algorithm;
[0042] The rejection mechanism includes: a debris positioning submodule that locates debris based on debris information from the image recognition submodule and calculates the time it takes for the debris to reach the rejection mechanism; a debris rejection submechanism that removes debris by air jet when the debris thrown in a parabola reaches the rejection position by controlling the solenoid valve 9; and a debris collection box 10 placed below the rejection position to receive the debris removed by the solenoid valve 9.
[0043] Tobacco leaf collection mechanism: After removing impurities, the tobacco leaves fall into the tobacco leaf collection mechanism 11 in a parabolic trajectory.
[0044] Preferably, the material transport module includes: a first conveyor belt 111 for receiving tobacco leaf material and a second conveyor belt 112 for spreading and accelerating the material; the end point of the first conveyor belt 111 is located above the starting point of the second conveyor belt 112, and the tobacco leaf material is thrown in a parabolic trajectory and falls into the starting point of the second conveyor belt.
[0045] Preferably, the material transport module further includes a fan structure 2 for attaching tobacco leaf material to the surface of the conveyor belt; the fan mechanism 2 includes: a first fan mechanism 21 that blows air onto the upper layer of tobacco leaf material on the second conveyor belt 112 to make the material evenly adhere to the conveyor belt; and a second fan mechanism 22 that blows parallel air towards the end point of the second conveyor belt 112 at the bottom end of the second conveyor belt 112 to make the tobacco leaf material move in a parabolic motion and form a solid-gas mixture containing the tobacco leaf material.
[0046] Preferably, the upper or lower imaging module includes: at least one set of parallel light sources 3: the width of one set of parallel light sources 3 is set according to the width of the tobacco leaves on the conveyor belt, that is, the width of the parabola after the tobacco leaves are thrown out; if two sets of parallel light sources 3 are selected, the two sets of parallel light sources 3 focus on the tobacco leaf imaging point to perform light concentration compensation on the tobacco leaves.
[0047] Preferably, the optical imaging module further includes: a glass assembly 4 for isolating the actuator of the photoelectric cleaning device and the tobacco leaves, wherein the slotted black background box 5 is embedded in the glass assembly corresponding to the branch, dividing the glass assembly into two parts; the glass assembly includes an upper glass assembly 41 and a lower glass assembly 42.
[0048] As shown in Figures 3 to 5, preferably, the grooved black background box includes: a U-shaped grooved structure 511, a beveled triangular structure 512, and a U-shaped bottom roller structure 513; the upper part of the grooved black background box can be provided with an opening for light to enter, namely a light slit 514 (5-12mm), and the bottom is provided with a debris outlet for falling debris or a rotation control mechanism for removing debris by rotating the roller.
[0049] In Figure 4, the upper part of the hypotenuse triangular structure 512 is provided with an air inlet 5121, and the lower part is provided with a dust and waste outlet 5122.
[0050] Preferably, in Figure 3, the U-shaped groove structure 511 further includes a light-blocking rod 5111 disposed at the focal point of the U-shaped arc, and in Figure 5, the U-shaped bottom roller structure 513 is provided with a rotatable light-blocking rod 5131.
[0051] Preferably, in the lower imaging module, the reflector 7 is placed vertically at the bottom of the lower imaging point, and the lower linear or area array camera 62 is placed at an angle to capture images of tobacco leaves in the reflector; the angle of the reflector 7 is adjustable.
[0052] Preferably, the lens of the linear or area array camera also includes a light-blocking plate 8.
[0053] Preferably, the image recognition submodule includes: an RGB color recognition algorithm unit that compares the colors of tobacco leaves and debris in an image based on RGB colors, calculates the probability of debris in the image, and identifies the debris using an RGB color recognition algorithm; an AI debris detection algorithm unit that trains an AI debris detection algorithm based on the debris image and detects whether debris exists in the tobacco leaf image based on the trained AI debris detection algorithm; and a debris information unit that locates the possible locations of debris based on the detection results of the RGB color recognition algorithm unit and the AI debris detection algorithm unit, and generates debris information.
[0054] Figures 6 and 7 show a comparison of the effects of the traditional optical imaging module and the optical imaging module of this device, respectively. It is found that the tobacco leaf images scanned by this device are clear, can clearly identify impurities, and have higher image quality.
[0055] Those skilled in the art will recognize that the units and method steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of the embodiments of the present invention.
Claims
1. An intelligent sheet smoke photoelectric impurity removal device, characterized in that, include: Material transport module: includes a conveyor belt for transporting tobacco leaf materials; Optical imaging modules: an upper imaging module that obtains tobacco leaf images by scanning the upper surface of the tobacco leaves with a camera, and a lower imaging module that obtains tobacco leaf images by scanning the lower surface of the tobacco leaves with a camera; the upper imaging module includes: a linear or area array camera that vertically captures images of the tobacco leaves below, a parallel light source that illuminates the upper imaging points of the tobacco leaf image from above for light compensation, and an upper slotted black background box for acquiring the black background of the upper tobacco leaf image; the lower imaging module includes: a linear or area array camera that vertically captures images of the tobacco leaves from the bottom, a parallel light source that illuminates the lower imaging points of the tobacco leaf image from below for light compensation, and a lower slotted black background box for acquiring the black background of the lower tobacco leaf image; Image recognition module: includes: an image acquisition submodule for acquiring tobacco leaf images captured by the upper imaging module and the lower imaging module, and an image recognition submodule for identifying debris information in the tobacco leaf images; the image recognition submodule includes identifying debris information based on RGB color recognition algorithm and AI debris detection algorithm; The rejection mechanism includes: a debris positioning submodule, a debris rejection submechanism, and a debris collection box; the debris positioning submodule locates the debris based on the debris information from the image recognition submodule and calculates the time it takes for the debris to reach the rejection mechanism; the debris rejection submechanism controls a solenoid valve to eject air and reject the debris when it reaches the rejection position after being thrown in a parabolic trajectory; the debris collection box is placed below the rejection position to receive the debris rejected by the solenoid valve. Tobacco leaf collection mechanism: The tobacco leaf collection mechanism collects tobacco leaves after removing impurities.
2. The intelligent sheet smoke photoelectric impurity removal device according to claim 1, characterized in that, The material transport module includes: a first conveyor belt for receiving tobacco leaf material and a second conveyor belt for spreading and accelerating the material; the end point of the first conveyor belt is located above the starting point of the second conveyor belt, and the tobacco leaf material is thrown in a parabolic trajectory and falls into the starting point of the second conveyor belt.
3. The intelligent sheet smoke photoelectric impurity removal device according to claim 2, characterized in that, The material transport module also includes a fan structure for attaching tobacco leaf material to the surface of the conveyor belt; the fan mechanism includes: a first fan mechanism that blows air onto the tobacco leaf material on the second conveyor belt to make the material evenly adhere to the conveyor belt; and a second fan mechanism that blows air in a direction perpendicular to the parabola of the tobacco leaf at the connection between the first and second conveyor belts to change the speed of the tobacco leaf.
4. The intelligent sheet smoke photoelectric impurity removal device according to claim 1, characterized in that, The optical imaging module includes: at least one set of parallel light sources: the number of parallel light sources is set according to the width of the tobacco leaves on the conveyor belt, that is, the parabolic width of the tobacco leaves after they are thrown out; if two sets of parallel light sources are selected, the two sets of parallel light sources are focused on the imaging point of the tobacco leaves to perform light concentration compensation on the tobacco leaves.
5. The intelligent sheet smoke photoelectric impurity removal device according to claim 1, characterized in that, The optical imaging module also includes a third fan structure that changes the parabolic angle of the tobacco leaves thrown from the material transport module as they move in a parabolic motion at a certain speed.
6. The intelligent sheet smoke photoelectric impurity removal device according to claim 1, characterized in that, The optical imaging module also includes a glass assembly for isolating the actuator of the photoelectric cleaning device and the tobacco leaves, wherein the slotted black background box is embedded in the glass assembly, dividing the glass assembly into two pieces.
7. The intelligent sheet smoke photoelectric impurity removal device according to claim 1, characterized in that, The slotted black background box includes: a U-shaped slotted structure, a beveled triangular structure, and a U-shaped bottom roller structure; the slotted black background box has an opening at the top for light to enter, and a debris outlet at the bottom for dropping debris or a rotation control mechanism for removing debris by rotating the roller.
8. An intelligent sheet smoke photoelectric impurity removal device according to claim 1 or 7, characterized in that, The U-shaped groove structure also includes a light-blocking rod placed at the focal point of the U-shaped arc.
9. An intelligent sheet smoke photoelectric impurity removal device according to claim 1 or 7, characterized in that, In the lower imaging module, a reflector is placed vertically at the bottom of the lower imaging point, and the linear or area array camera is placed at an angle to capture images of tobacco leaves within the reflector; the angles of the camera and the reflector are adjustable.
10. An intelligent sheet smoke photoelectric impurity removal device according to claim 1 or 7, characterized in that, The image recognition submodule includes: an RGB color recognition algorithm unit that compares the colors of tobacco leaves and debris in an image based on RGB colors, calculates the probability of debris in the image, and identifies the debris using an RGB color recognition algorithm; an AI debris detection algorithm unit that trains an AI debris detection algorithm based on the debris image and detects whether debris exists in the tobacco leaf image based on the trained AI debris detection algorithm; and a debris information unit that locates the possible locations of debris based on the detection results of the RGB color recognition algorithm unit and the AI debris detection algorithm unit, and generates debris information.