Cigarette surface droplet and defect identification method and identification device

Abstract

The invention relates to the technical field of tobacco production equipment, and discloses a cigarette surface droplet and defect identification method and identification device.The method comprises the following steps that S1, an air blowing mechanism is arranged at a detection station of a cigarette conveying path; s2, an air blowing mechanism is controlled to apply directional airflow to the surfaces of the cigarettes passing through the detection station; s3, collecting a first image of the surface of the cigarette before the air blowing mechanism is triggered; after the air blowing mechanism is triggered, collecting a second image of the surface of the cigarette; s4, acquiring a first area change rate of a change region between the first image and the second image; and S5, comparing the first area change rate with a preset first threshold value and a preset second threshold value, and then outputting corresponding judgment. Directional airflow is adopted as an intervention means, a pure physical mechanical mode is adopted, and potential electromagnetic interference of high-voltage static electricity on precise electronic equipment of a production line is thoroughly avoided.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of tobacco production equipment, and relates to a cigarette surface fly and defect identification method and identification device. BACKGROUND

[0002] In the high-speed cigarette rolling production process, the production speed of the rolling equipment is extremely high (usually not less than 8000 cigarettes per minute). In this environment, the rotation of the equipment drum easily causes tobacco fly to be generated and attached to the surface of the cigarette. These flies are non-quality defects, but their imaging features in the visual detection system are extremely similar to real cigarette defects (such as puncture, scratch, fly inclusion, etc.), which causes the traditional visual detection system relying on single static photography to be unable to effectively distinguish, resulting in a large number of intact cigarettes being misjudged as defective products and rejected. This not only directly leads to raw material waste and increased production cost, but also seriously restricts the further improvement of the production line efficiency.

[0003] To cope with the fly interference problem, the existing technology has adopted electrostatic adsorption technology. This scheme generates an electrostatic field to adsorb and remove the fly on the surface of the cigarette. However, the high-voltage electrostatic generating device may cause electromagnetic interference to the precise electronic sensing equipment (such as a high-speed camera) on the production line, affecting the stability and reliability of the detection system; the adsorption effect is significantly affected by the conductivity difference of the cigarette packaging materials (such as cigarette paper and tipping paper), resulting in unstable cleaning effect and poor compatibility. SUMMARY

[0004] The present application provides a cigarette surface fly and defect identification method and identification device, aiming to solve the above problems.

[0005] In one scheme, on the one hand, a cigarette surface fly and defect identification method is provided, comprising the following steps: S1, a gas blowing mechanism is arranged at a detection station of a cigarette conveying path; S2, the gas blowing mechanism is controlled to apply a directional gas flow to the surface of the cigarette passing through the detection station; S3, a first image of the surface of the cigarette is collected before the gas blowing mechanism is triggered, and a second image of the surface of the cigarette is collected after the gas blowing mechanism is triggered; S4, a first area change rate of the changed area between the first image and the second image is obtained; S5, the first area change rate is compared with a preset first threshold value and a second threshold value, if the first area change rate is less than the first threshold value, it is determined that the surface of the cigarette exists fly; if the first area change rate is greater than the second threshold value, it is determined that the surface of the cigarette exists defect; if the first area change rate is between the first threshold value and the second threshold value, a self-checking mode is started.

[0006] The "change region between the first image and the second image" refers to those image regions that have visible changes on the surface of the cigarette after the blowing action occurs due to the airflow intervention. The first area change rate refers to the percentage of the area of the change region relative to the area of the reference region, which is calculated after the first image (before blowing) and the second image (after the first blowing) are processed through feature matching and alignment.

[0007] The reference region area is the total area of all "suspected regions" in the first image. The "suspected region" refers to the total area of all regions that may be droplets or defects identified in the first image through preliminary image processing (such as threshold segmentation and edge detection).

[0008] In one scheme, the self-checking mode includes: obtaining the defect area in the second image and comparing it with the first set value, if the defect area in the second image is greater than the first set value, it is determined that the surface of the cigarette has defects; if the defect area in the second image is less than the first set value, the displacement and similarity between the defect area of the first image and the defect area of the second image are obtained; if the displacement is less than the second set value, it is determined that the surface of the cigarette has defects; if the displacement is greater than the second set value, the similarity is compared with the third set value, if the similarity is greater than the third set value, it is determined that the surface of the cigarette has defects; if the similarity is less than the third set value, it is determined that the surface of the cigarette has droplets.

[0009] Specifically, the defect area in the first image and the defect area in the second image are the areas of droplets or real defects.

[0010] In one scheme, the first set value is 150 pixels; the first set value is 10 pixels; and the first set value is 60%.

[0011] Specifically, the self-checking mode further includes: controlling the air blowing mechanism to apply directional airflow to the surface of the cigarette again, and collecting a third image; based on the change region between the second image and the third image, a second area change rate is calculated and obtained, and a determination is made according to the comparison result of the second area change rate with the first threshold and the second threshold.

[0012] Specifically, if the second area change rate is less than the first threshold, it is determined that the surface of the cigarette has defects; if the first area change rate is greater than the second threshold, it is determined that the surface of the cigarette has droplets; and if the first area change rate is between the first threshold and the second threshold, it is determined that the surface of the cigarette has defects.

[0013] In one scheme, the first image and the second image are subjected to feature matching and spatial alignment processing before step S4 is performed.

[0014] In one scheme, the feature matching and spatial alignment processing is performed by SIFT feature matching algorithm and affine transformation matrix to align the images and compensate for the displacement of the cigarette during the conveying process.

[0015] In one scheme, the first threshold A1 is 5%, and the second threshold A2 is 90%.

[0016] In one scheme, the other aspect provides a cigarette surface fly and defect identification device, comprising: A blowing mechanism is fixedly installed at the detection station of the cigarette conveying path, and the blowing mechanism is spaced apart from the cigarette. An image acquisition module includes two cameras, and the two cameras are spaced apart along the conveying direction of the cigarette, and the blowing mechanism is located between the two cameras. A processor, the blowing mechanism and the image acquisition module are electrically connected to the processor.

[0017] Specifically, a cigarette surface fly and defect identification device comprises two blowing mechanisms, and the two blowing mechanisms are fixedly installed at the detection station of the cigarette conveying path in sequence and are spaced apart from the cigarette. An image acquisition module includes three cameras, and the three cameras are spaced apart along the conveying direction of the cigarette, and the two blowing mechanisms are respectively located between the three cameras. A processor, each blowing mechanism and the image acquisition module are electrically connected to the processor.

[0018] The processor is configured to synchronously control the action timing of the blowing mechanism and the image acquisition module, receive and process the images transmitted by the image acquisition module, and perform image alignment, first area change rate calculation, second area change rate calculation and corresponding determination steps.

[0019] In one scheme, the blowing mechanism includes a blowing nozzle, an air pipe and a bracket; the blowing direction of the blowing nozzle points to the surface of the cigarette; the bracket is fixedly connected to the detection station of the cigarette conveying path; and the air pipe is fixedly connected to the blowing nozzle.

[0020] Specifically, an electromagnetic valve is installed on the air pipe to control the on-off of the air pipe, and the electromagnetic valve is electrically connected to the processor.

[0021] In one embodiment, the image acquisition module further includes multiple light sources, each of which is electrically connected to the processor; the multiple light sources are spaced apart along the cigarette conveying direction; and the two cameras are located between two adjacent light sources.

[0022] In one embodiment, the image acquisition module further includes a housing; the housing has multiple mounting holes; the multiple mounting holes are spaced apart along the cigarette conveying direction; multiple light sources and two cameras are sequentially and crosswise installed in the corresponding multiple mounting holes; the bracket is fixedly connected to the housing.

[0023] Specifically, the outer casing is fixedly connected to the inspection station on the cigarette conveying path by an external object.

[0024] The beneficial effects of this application are: By combining "airflow intervention" with "dual time-series image differential analysis", the fundamental problem of distinguishing between droplets and defects is solved by utilizing the essential physical property that droplets can be removed by airflow while defects cannot, thus greatly reducing the false rejection rate.

[0025] Compared to electrostatic adsorption technology, this invention uses directional airflow as the intervention method, which is a purely physical and mechanical approach. This completely avoids the potential electromagnetic interference of high-voltage static electricity on the precision electronic equipment of the production line, resulting in more stable system operation. Furthermore, its effectiveness is not affected by the conductivity of the cigarette material, and it has broader compatibility with various specifications of cigarette products.

[0026] By setting a first threshold and a second threshold, and by setting a self-checking mode, the system can handle difficult cases where the area change rate is in an ambiguous middle range. This three-level judgment mechanism (defect, droplet, re-inspection) significantly improves the intelligence and fault tolerance of the system's decision-making, effectively prevents misjudgments caused by uncertainty in a single detection, and further ensures detection accuracy.

[0027] The airflow intervention method does not require physical contact with the cigarette surface, so it will not cause any damage to the cigarette itself, nor will it interfere with the normal conveying and positioning of the cigarette on the drum, making it a perfect fit for high-speed and ultra-high-speed production lines. Attached Figure Description

[0028] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0029] Fig. 1 This is a schematic diagram of the workflow of a defect identification method according to an embodiment of this application; Fig. 2 is a processor control flow block diagram of an embodiment of the present application; Fig. 3 is a defect identification device structure schematic diagram of an embodiment of the present application; Fig. 4 is a distribution situation schematic diagram of a blowing mechanism, a camera and a light source on a shell of an embodiment of the present application; Wherein, 1, blowing mechanism; 11, blowing nozzle; 12, air pipe; 13, support; 14, electromagnetic valve; 2, image acquisition module; 21, camera; 22, light source; 23, shell; 3, processor; 4, cigarette. DETAILED DESCRIPTION

[0030] The specific embodiments of the present application will be further described in conjunction with the accompanying drawings and examples. The following examples are used to illustrate the present application, but not to limit the scope of the present application. Similarly, the following examples are only part of the embodiments of the present application, not all embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative efforts fall within the scope of the present application.

[0031] In the description of the present application, it should be understood that the terms "center", "longitudinal", "transverse", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "clockwise", "counterclockwise", "axial", "radial", "circumferential" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation of the present application.

[0032] In addition, the terms "first", "second" are only for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the technical features indicated. Therefore, the features defined with "first", "second" can explicitly or implicitly include at least one of the features. In the description of the present application, the meaning of "a plurality of" is at least two, for example, two, three, etc., unless otherwise specifically limited.

[0033] In the present application, unless specifically defined and limited otherwise, the terms "mount", "connect", "connection", "fixed", and the like should be understood broadly, for example, can be fixed connection, can also be detachable connection, or integrated; can be mechanical connection, or electrical connection or communication with each other; can be directly connected, or indirectly connected through an intermediate medium, or the internal communication of two elements or the interaction relationship between two elements, unless otherwise specifically limited. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances.

[0034] In the present application, unless specifically defined and limited otherwise, the first feature is "on" or "under" the second feature. The first and second features can be in direct contact, or the first and second features can be in indirect contact through an intermediate medium. Moreover, the first feature "above", "over" and "on" the second feature can be that the first feature is directly above or obliquely above the second feature, or only indicates that the horizontal height of the first feature is higher than that of the second feature. The first feature "below", "under" and "under" the second feature can be that the first feature is directly below or obliquely below the second feature, or only indicates that the horizontal height of the first feature is less than that of the second feature.

[0035] In the present application, the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" mean that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. In the present specification, the illustrative description of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any appropriate manner in any one or more embodiments or examples. In addition, those skilled in the art can combine and combine different embodiments or examples described in the specification and the features of different embodiments or examples without contradiction.

[0036] The present application makes improvements and innovations, and proposes the following embodiments.

[0037] In some embodiments, referring to Figs. 1 to 4 In one aspect, a cigarette surface fly and defect identification method is provided, comprising the following steps: S1, a gas blowing mechanism 1 is arranged at the detection station of the cigarette 4 conveying path; S2, control the gas blowing mechanism 1 to apply directional airflow to the surface of the cigarette 4 passing through the detection station; S3, before the gas blowing mechanism 1 is triggered, the first image of the surface of the cigarette 4 is collected; after the gas blowing mechanism 1 is triggered, the second image of the surface of the cigarette 4 is collected; S4, obtaining the first area change rate of the change area between the first image and the second image; S5, compare the first area change rate with the preset first threshold and second threshold, if the first area change rate is less than the first threshold, it is determined that the flying spit exists on the surface of the cigarette 4; if the first area change rate is greater than the second threshold, it is determined that the defect exists on the surface of the cigarette 4; if the first area change rate is between the first threshold and the second threshold, the self-checking mode is started.

[0038] Wherein, the change area between the first image and the second image refers to those image areas which have visible changes on the surface of the cigarette 4 due to the airflow intervention after the blowing action occurs. The first area change rate refers to the percentage of the change area relative to the reference area after the first image (before blowing) and the second image (after blowing for the first time) are processed through feature matching and alignment.

[0039] The reference area is the total area of all "suspicious areas" in the first image. The so-called "suspicious area" refers to the total area of all areas which may be flying spit or defect identified in the first image through preliminary image processing (such as threshold segmentation, edge detection).

[0040] By combining "airflow intervention" with "double time sequence image difference analysis", the essential physical characteristics that flying spit can be removed by airflow and defect cannot be removed are utilized, so that the difficult problem of distinguishing flying spit from defect is fundamentally solved, and the false rejection rate is greatly reduced.

[0041] Compared with the electrostatic adsorption technology, the present application uses directional airflow as the intervention means, which belongs to a pure physical and mechanical way, and completely avoids the potential electromagnetic interference of high-voltage static electricity on the precision electronic equipment of the production line, so that the system runs more stably. At the same time, its effect is not affected by the conductivity of the cigarette 4 material, and it has more extensive compatibility for cigarette 4 products of various specifications.

[0042] By setting the first threshold and the second threshold, and setting the self-checking mode, the system can handle difficult cases with an intermediate ambiguous area. This three-level judgment (defect, flying spit, recheck) mechanism significantly improves the intelligence and fault tolerance of system decision-making, effectively prevents misjudgment caused by single detection uncertainty, and further ensures the detection accuracy.

[0043] The airflow intervention method does not need to have physical contact with the surface of the cigarette 4, which will not cause any damage to the cigarette 4 itself, nor will it interfere with the normal conveying and positioning of the cigarette 4 on the drum, and it is perfectly suitable for high-speed and ultra-high-speed production lines.

[0044] In one embodiment, the self-inspection mode comprises: obtaining the defect area in the second image and comparing it with the first set value, if the defect area in the second image is greater than the first set value, it is determined that the surface of the cigarette exists defects; if the defect area in the second image is less than the first set value, the displacement amount and the similarity between the defect area of the first image and the defect area of the second image are obtained; if the displacement amount is less than the second set value, it is determined that the surface of the cigarette exists defects; if the displacement amount is greater than the second set value, the similarity is compared with the third set value, if the similarity is greater than the third set value, it is determined that the surface of the cigarette exists defects; if the similarity is less than the third set value, it is determined that the surface of the cigarette exists fly ash.

[0045] Specifically, the defect area in the first image and the defect area in the second image are the areas of fly ash or real defects.

[0046] When the area change rate is in the fuzzy interval, the self-inspection mode introduces the features of the defect area, the displacement amount and the image similarity for comprehensive decision-making. This multi-feature fusion decision-making model greatly improves the scientificity and accuracy of the determination of difficult cases.

[0047] It can accurately distinguish between "blown fly ash" and "static defect": the position of the real defect is stable before and after blowing. By calculating the displacement amount, it can easily identify those fly ash that is blown by the air flow but not completely removed (large displacement amount), so as to distinguish it from the real defect (small displacement amount). Similarity comparison further provides evidence from the image texture level, forming a strong cross-validation.

[0048] For some fly ash that is attached tightly or not completely blown away due to static electricity, but moves as a whole, the simple area change rate may fail. However, by analyzing its position movement (displacement amount) and shape retention (high similarity), the system can still accurately identify it as fly ash, not a defect, thereby effectively preventing the missed judgment of such special fly ash.

[0049] The judgment process is designed ingeniously and closely linked. First, it is screened by area size, then it is judged again by displacement amount, and finally it is judged by similarity. This ladder-type, funnel-type judgment structure ensures the rigor and efficiency of the decision-making process, and the calculation resources are concentrated on the most needed cases.

[0050] In one embodiment, the first set value is 150 pixels; the first set value is 10 pixels; and the first set value is 60%.

[0051] The abstract decision logic is converted into concrete, quantified numerical thresholds, making the intelligent algorithm finally land in the implementation of the industrial control system. For example, "the first set value 150 pixels" defines the minimum size of "significant defects" that need to start fine judgment.

[0052] These thresholds (such as a displacement tolerance of 10 pixels and a similarity threshold of 60%) are based on a balance point set by a large amount of experimental data. They can effectively filter out interference caused by image noise, minor alignment errors, etc. while effectively capturing real droplet movement and morphological changes, thereby maintaining a low false positive rate while ensuring high detection rate.

[0053] The reference parameter values are explicitly given, providing a clear reference starting point for production line engineers to fine-tune and optimize for different product specifications (such as different sizes of cigarettes) or specific working conditions, ensuring that the technical solution has good portability and adaptability.

[0054] In one embodiment, the self-checking mode includes: The air blowing mechanism 1 applies a directional air flow to the surface of the cigarette 4 again, and a third image is collected; Based on the change area between the second image and the third image, a second area change rate is calculated, and a judgment is made according to the comparison result of the second area change rate with the first threshold and the second threshold.

[0055] When the result of the first detection is in the ambiguous interval, the system does not arbitrarily make a final judgment of "qualified" or "defective", but obtains new evidence through secondary intervention, greatly reducing the risk of misjudgment caused by single accidental factors (such as tight droplet adhesion, new droplet instantaneous adhesion, etc.).

[0056] It is particularly suitable for scenarios where droplets partially obscure real defects. The first air blowing may only blow away part of the droplets, resulting in an ambiguous change rate; the second air blowing can further remove the remaining droplets, making the obscured defects fully exposed and accurately captured in the second difference analysis.

[0057] This mechanism enables the system to have self-verification and error correction capabilities, upgrading from an open-loop, one-time judgment to a closed-loop, iterative intelligent decision-making process, significantly improving the fault tolerance and intelligence level of the entire system.

[0058] In one embodiment, if the second area change rate is less than the first threshold, it is determined that the surface of the cigarette 4 has a defect; if the first area change rate is greater than the second threshold, it is determined that the surface of the cigarette 4 has a droplet; if the first area change rate is between the first threshold and the second threshold, it is determined that the surface of the cigarette 4 has a defect.

[0059] When the re-inspection still cannot distinguish the defect, the final judgment is made as a defect. This follows the core principle of "better to kill than to let go" in industrial quality inspection, ensuring the reliability of the quality of the products leaving the factory, and avoiding risky cigarettes 4 flowing into the market. The rule sets a clear endpoint for the self-inspection process, preventing the system from getting stuck in an infinite loop of detection, ensuring the efficiency and decisiveness of the production line decision, while guaranteeing the quality bottom line.

[0060] In one embodiment, before performing step S4, the first image and the second image are subjected to feature matching and spatial alignment processing.

[0061] In one embodiment, the feature matching and spatial alignment processing aligns the images through a SIFT feature matching algorithm and an affine transformation matrix, compensating for the displacement of the cigarettes 4 during the conveying process.

[0062] Through the SIFT algorithm and the affine transformation, the displacement and slight rotation of the cigarettes 4 during high-speed conveying are accurately compensated for, ensuring that the differential analysis is for the same physical point on the cigarettes 4. This prevents normal displacement of the cigarettes 4 from being misjudged as a "change area".

[0063] Performing the differential on the aligned images makes the stable background and the real defects perfectly canceled out, while the true change signal (flying ash) is highlighted, greatly improving the signal-to-noise ratio and accuracy of the "area change rate" calculation.

[0064] Specifically, the SIFT feature matching algorithm and the affine transformation matrix are both mature image processing technologies on the market. The specific working steps are as follows: First, SIFT feature extraction of the first image and the second image; The system independently runs the SIFT algorithm on the first image (before blowing) and the second image (after blowing).

[0065] The SIFT (Scale-Invariant Feature Transform) algorithm finds some stable and unique points in the image, such as corner points, edge highlight centers, etc. These points can be repeatedly detected even if the image is moved, rotated, or scaled. In the cigarette 4 image, these points may be small texture points on the cigarette paper, corners of the trademark printing, etc.

[0066] For each detected key point, SIFT calculates a 128-dimensional feature vector called a descriptor. This descriptor describes the gradient direction distribution of the pixel region around the key point. As long as the region does not change, its descriptor remains essentially unchanged.

[0067] Second, feature matching of the first image and the second image; All the keypoint descriptors in the first image are compared with all the keypoint descriptors in the second image to find the "best match" for each keypoint in the first image.

[0068] A k-Nearest Neighbor algorithm is usually used to speed up the matching process.

[0069] To eliminate false matches (e.g. two different texture points happen to look "similar"), Lowe's Ratio Test is used. In simple terms, only the matches where the "best match" is much better than the "second best match" are kept. This filters out a set of high-quality, reliable matches.

[0070] Compute the affine transformation matrix between the first and second image. Using the high-quality matches from the previous step, estimate a mathematical model that describes the geometric transformation between the two images.

[0071] Assume that the motion of cigarette 4 on the drum is an affine transformation, which can include translation, rotation, scaling and shearing. This is a good approximation of the cigarette motion on a high-speed production line.

[0072] Solve for an optimal affine transformation matrix M using the RANSAC algorithm.

[0073] The advantage of RANSAC is that it is very robust to false matches. Even if a few matches are wrong, it can still estimate the correct transformation matrix. The process is: Randomly select a minimal set of matches (3 matches for affine transformation).

[0074] Compute a transformation matrix M.

[0075] Test all the other matches using this M to see how many matches fit the model (these matches are called "inliers").

[0076] Repeat many times and finally adopt the transformation matrix M that has the most "inliers".

[0077] Third, align the first and second images in image space. Apply the affine transformation matrix M computed in the second step to the second image to "warp" it so that it is aligned with the first image in space.

[0078] Use the affine transformation to map each pixel in the second image to a new location according to the matrix M.

[0079] After this operation, the position and posture of the cigarette 4 which has been displaced and rotated in the second image will become almost completely consistent with the cigarette 4 in the first image.

[0080] The SIFT feature is invariant to rotation, scale and brightness change, and is very suitable for the complex working conditions of cigarette 4 images in high-speed production lines. Combined with the RANSAC algorithm, it has strong fault tolerance for incorrect feature matching, and can still estimate a high-precision transformation matrix from data that may have incorrect matching, ensuring the stability and reliability of the system in actual industrial environments.

[0081] In one embodiment, the first threshold A1 is 5%, and the second threshold A2 is 90%.

[0082] The lower limit of 5% can effectively filter out minor changes caused by image noise or minor alignment errors, avoiding misjudgment of minor defects as fly ash; the upper limit of 90% ensures that only objects that are almost completely removed are judged as fly ash, avoiding misjudgment of partially detached defects as fly ash. This threshold range provides a clear and efficient boundary for the identification of most fly ash and defects, while retaining a reasonable fuzzy interval to trigger a more cautious re-inspection mechanism, which is the key to achieving a balance between efficiency and accuracy.

[0083] In one embodiment, the other aspect provides a cigarette surface fly ash and defect identification device, comprising: The air blowing mechanism 1 is fixedly installed at the detection station of the cigarette 4 conveying path, and the air blowing mechanism 1 is spaced apart from the cigarette 4; The image acquisition module 2 includes two cameras 21, which are spaced apart along the conveying direction of the cigarette 4, and the air blowing mechanism 1 is located at the interval between the two cameras 21; The processor 3 is electrically connected to the air blowing mechanism 1 and the image acquisition module 2.

[0084] The "blowing intervention" and "front and rear image acquisition" functions are integrated into one, and through the synchronous control of the processor 3, a complete "perception-intervention-re-perception-decision" automated detection closed loop is formed, providing a solid hardware foundation for the implementation of the core method.

[0085] The "camera-blowing-camera" spacing arrangement method enables the series of actions of "taking the first photo, triggering blowing, and taking the second photo" to be completed in sequence with extremely high timing accuracy in the continuous process of cigarette conveying. This layout logic is clear, maximizes the space occupation of the detection station, and ensures the continuity and accuracy of action execution.

[0086] The hardware architecture is designed for high-speed pipeline. The cooperation of double cameras and blowing mechanism ensures that the system can complete all data acquisition and intervention actions in a very short time, perfectly adapting to the high-speed production rhythm of more than 8000 cigarettes per minute.

[0087] Specifically, the cigarette surface fly ash and defect identification device further comprises: two blowing mechanisms 1, two blowing mechanisms 1 are sequentially and fixedly installed at the detection station of the cigarette 4 conveying path, and each blowing mechanism 1 is arranged at intervals with the cigarette 4; An image acquisition module 2 comprises three cameras 21, three cameras 21 are sequentially and spacedly arranged along the conveying direction of the cigarette 4, and two blowing mechanisms 1 are respectively located at two intervals of three cameras 21; A processor 3, each blowing mechanism 1 and the image acquisition module 2 are electrically connected with the processor 3.

[0088] Among them, the processor 3 is configured to: synchronously control the action timing of the blowing mechanism 1 and the image acquisition module 2; receive and process the images transmitted by the image acquisition module 2, and perform image alignment, first area change rate, second area change rate calculation and corresponding determination steps.

[0089] By arranging two blowing mechanisms 1 and three cameras 21, the first camera 21, the first blowing mechanism 1, the second camera 21, the second blowing mechanism 1 and the third camera 21 are sequentially distributed along the conveying direction of the cigarette 4; the first blowing mechanism 1, the first camera 21 and the second camera 21 are used to detect fly ash and defects on the cigarette 4 (primary detection), and the second blowing mechanism 1 and the third camera 21 establish a physically independent special re-detection channel for the "self-detection mode". This makes the "primary detection" and "re-detection" processes of the system can be separated in space and connected in time, realizing the process and parallelization of detection operation. When the current cigarette 4 enters the re-detection stage, the next cigarette 4 can immediately start primary detection, greatly optimizing the detection timing, minimizing the impact on the production line rhythm, and being especially suitable for ultra-high-speed production scenes.

[0090] The special re-detection camera 21 can image the cigarette 4 again after the first blowing intervention, and perform differential analysis with the second image. This is equivalent to providing a second and more targeted diagnosis opportunity for those "difficult and complex" cases. By obtaining the new key evidence of the second area change rate, the system can make a more reliable and accurate final determination, significantly reducing the risk of misjudgment in ambiguous cases.

[0091] The architecture solidifies the "three-level judgment mechanism" of the method in hardware, and makes the system have stronger self-verification and error correction capability. When a cigarette 4 is judged to need re-inspection, it is automatically guided to the next detection node for review, the whole process does not need manual intervention, forming a highly automated and intelligent closed-loop quality control system, and the overall fault tolerance and intelligence level of the system are greatly improved.

[0092] The independent air blowing mechanism 1 is provided for re-inspection, which can optimize the blowing pressure, angle and duration for re-inspection, and ensure the best effect and consistency of air flow intervention. This avoids the problem of unstable pressure caused by rapid and continuous triggering of the same air blowing mechanism 1, and ensures that the intervention conditions of the two links of initial inspection and re-inspection are in a controlled and ideal state.

[0093] In an embodiment, the air blowing mechanism 1 includes a blowing nozzle 11, an air pipe 12 and a bracket 13; the air outlet direction of the blowing nozzle 11 points to the surface of the cigarette 4; the bracket 13 is fixedly connected at the detection station of the cigarette 4 conveying path; and the air pipe 12 is fixedly connected with the blowing nozzle 11.

[0094] Specifically, the air pipe 12 is provided with an electromagnetic valve 14 for controlling the on-off of the air pipe 12, and the electromagnetic valve 14 is electrically connected with the processor 3.

[0095] The blowing nozzle 11 is fixed by the bracket 13, which ensures that the air flow direction is always accurately pointed to the imaging area, forming a stable and controllable air flow intervention field.

[0096] In an embodiment, the image acquisition module 2 further includes a plurality of light sources 22, each of which is electrically connected with the processor 3; the plurality of light sources 22 are arranged at intervals along the conveying direction of the cigarette 4; and the three cameras 21 are respectively located between two adjacent light sources 22.

[0097] The plurality of light sources 22 and the cameras 21 are arranged in cross intervals, which can effectively eliminate the reflection and shadow on the surface of the cigarette 4, and provide a uniform and stable lighting environment for the high-speed camera 21, which is a prerequisite for obtaining high-quality and analyzable images.

[0098] In an embodiment, the image acquisition module 2 further includes a housing 23; a plurality of mounting holes are formed in the housing 23; the plurality of mounting holes are arranged at intervals along the conveying direction of the cigarette 4; the plurality of light sources 22 and the three cameras 21 are sequentially and crossly mounted in the corresponding plurality of mounting holes; and the bracket 13 is fixedly connected with the housing 23.

[0099] The integrated housing 23 design integrates the light sources 22, the cameras 21 and the blowing bracket 13 into one, forming an independent detection module. The module has a compact structure, which is convenient for rapid installation, debugging and maintenance on the existing rolling equipment, greatly improving the engineering application value and promotion convenience of the technology.

[0100] Specifically, the shell 23 is fixedly connected with the outer object at a detection station of the cigarette 4 conveying path.

[0101] The above is only an optional embodiment of the present application, and is not used to limit the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application. Although the embodiments of the present application have been shown and described above, it should be understood that the above-described embodiments are exemplary and should not be construed as limiting the present application. Those skilled in the art can make changes, modifications, replacements and variations to the above-described embodiments within the scope of the present application.

Claims

1. A method for identifying surface fly and defects of a cigarette, characterized in that, The method comprises the following steps: S1, a blowing mechanism is arranged at a detection station of a cigarette conveying path; S2, the blowing mechanism is controlled to apply a directional air flow to the surface of a cigarette passing through the detection station; S3, a first image of the surface of the cigarette is captured before the blowing mechanism is triggered, and a second image of the surface of the cigarette is captured after the blowing mechanism is triggered; S4, a first area change rate of a changed area between the first image and the second image is obtained; S5, the first area change rate is compared with a first threshold value and a second threshold value, if the first area change rate is less than the first threshold value, it is determined that the surface of the cigarette has droplets, if the first area change rate is greater than the second threshold value, it is determined that the surface of the cigarette has defects, and if the first area change rate is between the first threshold value and the second threshold value, a self-checking mode is started.

2. The identification method according to claim 1, characterized in that, The self-checking mode comprises: comparing a defect area in the second image with a first set value, if the defect area in the second image is greater than the first set value, it is determined that the surface of the cigarette has defects, if the defect area in the second image is less than the first set value, a displacement and a similarity between a defect area of the first image and a defect area of the second image are obtained, if the displacement is less than a second set value, it is determined that the surface of the cigarette has defects, if the displacement is greater than the second set value, the similarity is compared with a third set value, if the similarity is greater than the third set value, it is determined that the surface of the cigarette has defects, and if the similarity is less than the third set value, it is determined that the surface of the cigarette has droplets.

3. The identification method according to claim 2, characterized in that, The first set value is 150 pixels, the first set value is 10 pixels, and the first set value is 60%.

4. The identification method according to claim 3, characterized in that, Before step S4 is performed, the first image and the second image are subjected to feature matching and spatial alignment processing.

5. The identification method according to claim 4, characterized in that, The feature matching and spatial alignment processing align images through a SIFT feature matching algorithm and an affine transformation matrix, and compensate for displacement of the cigarette in the conveying process.

6. The identification method according to any one of claims 1 to 5, characterized in that, The first threshold value A1 is 5%, and the second threshold value A2 is 90%.

7. A cigarette surface fly and defect recognition device, based on the recognition method according to any one of claims 1-6, characterized in that, It comprises: A blowing mechanism is fixedly installed at a detection station of a cigarette conveying path, and the blowing mechanism is arranged at a distance from the cigarette. An image acquisition module comprises two cameras, the two cameras are arranged at a distance along the conveying direction of the cigarette, and the blowing mechanism is located at the distance between the two cameras. A processor, the blowing mechanism and the image acquisition module are electrically connected to the processor.

8. The identification method according to claim 7, characterized in that, The blowing mechanism comprises a blowing nozzle, an air pipe and a bracket, the blowing direction of the blowing nozzle points to the surface of the cigarette, the bracket is fixedly connected at the detection station of the cigarette conveying path, and the air pipe is fixedly connected with the blowing nozzle.

9. The identification method according to claim 8, characterized in that, The image acquisition module further comprises a plurality of light sources, each light source is electrically connected to the processor, a plurality of light sources are arranged at a distance along the conveying direction of the cigarette, and the two cameras are respectively located between two adjacent light sources.

10. The identification method according to claim 9, characterized in that, The image acquisition module further comprises a shell; a plurality of mounting holes are formed in the shell; the plurality of mounting holes are arranged at intervals along the cigarette conveying direction; the plurality of light sources and the two cameras are installed in the corresponding plurality of mounting holes in sequence and cross each other; and the support is fixedly connected with the shell.

Images