Labeling bubble identification method and system

Abstract

This application discloses a method and system for identifying bubble structures in a label cap; belonging to the field of defect detection technology, the method includes: acquiring a label image of the label cap, identifying mutually separated candidate bubble regions and determining their spatial proximity; applying a local recoverable pressure perturbation to the target bubble region, acquiring the response label images during and after the pressure perturbation; determining the morphological changes of the target bubble region and adjacent candidate bubble regions, and determining a connected bubble group connected through a gas channel below the label based on the linkage response relationship between the two with the pressure perturbation; generating a label bubble identification result based on the overall coverage of the connected bubble group, and determining the label cap's adhesion quality status; this application can identify mutually separated but actually gas-connected bubble structures in a two-dimensional image.

Description

Technical Field

[0001] This invention belongs to the field of defect detection technology, specifically a method and system for identifying air bubbles in labels. Background Technology

[0002] During the labeling, pressing, and curing process, air bubbles may form between the label and the cap surface due to factors such as insufficient local adhesion of the label, uneven adhesive layer distribution, or residual gas between the label and the cap. To avoid air bubbles affecting label adhesion quality, existing detection methods typically use industrial cameras to acquire images of the label surface, identify suspected bulge areas based on local brightness, texture, edge, or contour changes, obtain corresponding bubble areas through connected component segmentation, and then judge the adhesion quality based on parameters such as the area, size, or location of each bubble area.

[0003] However, the gas space formed between the label and the cap does not necessarily correspond one-to-one with the visible bubble outlines on the label surface. When there is a gas channel with a low height or narrow width between two locally bulging areas, the deformation of the label surface corresponding to the gas channel may not be sufficient to meet the image recognition conditions. This causes two bubbles that actually belong to the same gas-connected structure to appear as multiple separate abnormal regions in the two-dimensional label image. Existing detection methods based on static image connected components can usually only identify each bubble region based on the visible abnormal outlines, making it difficult to further determine whether these separate bubble regions have an actual gas connection relationship under the label.

[0004] In this case, if the quality is evaluated separately based on the area or span of each visible bubble region, multiple bubble regions whose individual dimensions do not exceed the allowable range, but which actually constitute a large gas communication structure, may be judged as permissible defects, resulting in the test results not accurately reflecting the actual range of the bubble structure between the label and the cap.

[0005] Therefore, the main problem with existing label bubble detection technology is that it is difficult to identify visually separated bubble areas that are actually connected under the label based solely on a two-dimensional static image of the label surface, which in turn affects the accurate judgment of the overall range of label bubble defects and the bonding quality. Summary of the Invention

[0006] To address the problem in existing label bubble detection processes where relying solely on two-dimensional static images of the label surface makes it difficult to identify visually separated but actually connected bubble regions beneath the label, thus affecting the overall extent of bubble defects and the assessment of adhesion quality, this application provides a label bubble identification method and system. By applying local recoverable pressure perturbation to visually separated candidate bubble regions, the gas connectivity state beneath the label, which is difficult to observe directly, is transformed into a dynamic morphological linkage relationship between different candidate bubble regions that can be characterized by the response label image. Based on this, connected bubble groups are formed and group-level quality evaluation is performed.

[0007] In a first aspect, this application provides a label bubble recognition method for a label cap, comprising: acquiring a label image of a label cap; identifying multiple mutually separated candidate bubble regions; and determining spatial proximity relationships based on the spatial positions of each candidate bubble region; applying a local recoverable pressure perturbation to a target bubble region among the candidate bubble regions; and acquiring a response label image during the pressure perturbation and after the pressure is released; determining the morphological changes of the target bubble region based on the response label image; filtering other candidate bubble regions spatially separated from the target bubble region based on the spatial proximity relationships; and determining the corresponding morphological changes of the filtered other candidate bubble regions; determining a visually separated but connected bubble group through a gas channel below the label based on the linkage response relationship between the target bubble region and the other candidate bubble regions generated by the pressure perturbation; generating a label bubble recognition result based on the overall coverage of the connected bubble group; and determining the adhesion quality state of the label cap based on the label bubble recognition result.

[0008] The above method no longer judges the bubble structure solely based on the two-dimensional connectivity of candidate bubble regions in a single label image. Instead, it utilizes local recoverable pressure perturbation to redistribute the gas beneath the label and characterizes the hidden gas connectivity relationship through the linkage morphological changes between the target bubble region and the spatially separated candidate bubble regions. This allows multiple visually separated regions that actually belong to the same gas connectivity structure to be grouped into a connected bubble group, and the overall coverage of the connected bubble group is used for quality evaluation. This reduces the possibility of missing overall bubble defects due to the separate evaluation of multiple visually separated small bubbles.

[0009] In one possible implementation, the label image is segmented into bubble regions to obtain multiple unconnected candidate bubble regions, and the region boundary of each candidate bubble region is determined; the minimum boundary distance between the region boundaries of any two candidate bubble regions is calculated, and the spatial proximity relationship is established between two candidate bubble regions whose minimum boundary distance is not greater than a preset spatial proximity threshold.

[0010] Through the above implementation method, before performing pressure perturbation, multiple visually separated candidate bubble regions are first screened for proximity using the region boundaries in the two-dimensional image and the actual spatial distance. This allows subsequent pressure perturbation and linkage response judgments to focus on candidate bubble region pairs that have the possibility of hiding gas channels, reducing unnecessary connectivity verification for candidate bubble regions that are far apart.

[0011] In one possible implementation, the target bubble region is determined from candidate bubble regions that have at least one spatially adjacent candidate bubble region and whose connectivity determination has not yet been completed; pressure is applied to the target bubble region according to the calibrated recoverable pressure range and a preset disturbance duration is maintained; the response tag image is acquired according to a preset acquisition sequence during the recovery period before pressure is applied, during pressure application, and after pressure is released, forming a disturbance response image sequence corresponding to the target bubble region.

[0012] Through the above implementation method, the pressure disturbance is limited to the calibrated recoverable pressure range, and continuous response tag images are acquired simultaneously before pressure is applied, during the pressure application stage, and after pressure is released. This allows the pressure disturbance to produce morphological changes sufficient to be detected by the image, while reducing the possibility that excessive pressure may cause the tag morphology to become unrecoverable and affect subsequent identification results. Furthermore, by forming a disturbance response image sequence with a clear acquisition time sequence, a temporal data basis is provided for subsequently determining the sequential relationship between contraction, expansion, and recovery responses.

[0013] In one possible implementation, each response label image in the disturbance response image sequence is registered with the label image before pressure is applied; at least one of the area, equivalent size, and boundary position of the target bubble region and the remaining candidate bubble regions that have the spatial proximity relationship with the target bubble region and are separated from each other in the label image is extracted at each acquisition time; and the corresponding morphological change is determined according to the change of the corresponding parameter at each acquisition time relative to before pressure is applied.

[0014] Through the above implementation method, the influence of label cover position offset, image translation or imaging direction change on the comparison of regional morphology during the detection process is reduced by position registration. The morphological parameters that can be quantified, such as regional area, equivalent size and boundary position, are used to describe the changes of each candidate bubble region over time. This allows the morphological response at different acquisition times to be compared under a unified position reference, providing comparable data objects for judging whether there is a real linkage response between candidate bubble regions.

[0015] In one possible implementation, the contraction response of the target bubble region and the expansion response of the remaining candidate bubble regions are determined during the pressure application phase, and the recovery response of the target bubble region and the reverse recovery response of the remaining candidate bubble regions are determined after the pressure is released; the pressure response time difference between the contraction response and the expansion response, and the recovery response time difference between the recovery response and the reverse recovery response are determined; when both the pressure response time difference and the recovery response time difference are within the corresponding allowable time range, the linkage response intensity is determined based on the amplitude of each response, and the gas connectivity relationship is determined based on the linkage response intensity.

[0016] By implementing the above methods, the determination of gas connectivity is based on the "contraction-expansion" correspondence during the pressure application phase and the "recovery-reverse recovery" correspondence during the pressure release phase. The determination is made in conjunction with the response time difference and response amplitude. This makes it less likely that accidental changes in candidate bubble regions that are spatially adjacent but do not have gas connectivity due to mechanical vibration, overall elastic deformation of the label, or fluctuations in image detection will be directly identified as gas connectivity, thereby improving the reliability of the determination of hidden gas connectivity.

[0017] In one possible implementation, after determining that any of the target bubble regions is connected to at least one of the remaining candidate bubble regions through the gas channel below the label, each candidate bubble region is taken as a member node, and the candidate bubble region pairs that are determined to have a gas connection relationship are taken as connecting edges to construct a bubble connection relationship. All member nodes that are directly or indirectly connected through the connecting edges are merged into the same connected bubble group.

[0018] The above implementation method does not require that any two candidate bubble regions in the same actual gas connectivity structure establish a direct connection relationship. Instead, it uses the gas connectivity relationship that has been verified to build the connection relationship between member nodes. For example, when the first candidate bubble region is connected to the second candidate bubble region, or the second candidate bubble region is connected to the third candidate bubble region, the three candidate bubble regions can be uniformly merged into the same connected bubble group, thereby representing the overall gas connectivity structure composed of multiple visually separated regions.

[0019] In one possible implementation, generating a tag bubble recognition result based on the overall coverage of the connected bubble group includes: obtaining the regional boundaries of each candidate bubble region within the connected bubble group; determining a group-level outer envelope region surrounding each member candidate bubble region based on all the regional boundaries; determining the total area of ​​visible bubbles at the group level based on the sum of the areas of each member candidate bubble region; and determining the maximum span at the group level based on the maximum actual distance between each member candidate bubble region; and recording the group-level outer envelope region, the total area of ​​visible bubbles at the group level, and the maximum span at the group level in the tag bubble recognition result.

[0020] Through the above implementation method, the connected bubble group is characterized from three aspects: the overall spatial range, the actual visible bubble area, and the spatial span between member candidate bubble regions. This eliminates the need to manually estimate the hidden gas channels that cannot be directly observed in the label image as bubble areas. At the same time, the overall defect range formed by multiple member candidate bubble regions can still be reflected by the group-level outer envelope region and the group-level maximum span, thereby providing evaluation parameters that can be directly determined from image data for subsequent group-level quality evaluation.

[0021] In one possible implementation, determining the bonding quality status of the label cap based on the label bubble identification result includes: comparing the total area of ​​visible bubbles at the group level and the maximum span at the group level of each of the connected bubble groups with the allowable threshold for bubble area and the allowable threshold for bubble span, respectively; and performing corresponding comparisons on candidate bubble regions not included in the connected bubble groups according to their area and maximum span; when the total area of ​​visible bubbles at the group level or the maximum span at the group level of any of the connected bubble groups exceeds the corresponding allowable threshold, or when the area area or the maximum span of any candidate bubble region not included in the connected bubble group exceeds the corresponding allowable threshold, the bonding quality status of the label cap for label bubble defects is determined to be unqualified; otherwise, it is determined to be qualified.

[0022] Through the above implementation method, group-level evaluation is adopted for candidate bubble regions that have been identified as having hidden gas connectivity, while individual evaluation is adopted for candidate bubble regions that have not formed gas connectivity. This allows the quality judgment object to match the actual bubble structure type, reducing the situation where gas connectivity structures where the size of multiple individual units does not exceed the allowable threshold, but the overall range after combination exceeds the allowable range, are judged as qualified.

[0023] In one possible implementation, after determining that the target bubble region and any of the other candidate bubble regions have a gas communication relationship based on the linkage response intensity, the other candidate bubble regions are treated as new target bubble regions and subjected to local recoverable pressure disturbances. When the original target bubble region generates an expansion response during the pressure-bearing phase of the new target bubble region and generates a reverse recovery response after the pressure is released, and the corresponding pressure response time difference and recovery response time difference are both within the corresponding allowable time range, and the linkage response intensity determined based on the amplitude of each response reaches a preset linkage response intensity threshold, the two candidate bubble regions are determined to be connected through the gas channel below the label.

[0024] Through the above implementation method, after the gas connectivity relationship is obtained by the initial pressure disturbance, the reverse pressure disturbance is performed again by exchanging the pressure object, so that the gas migration response between the two candidate bubble regions can be verified from both disturbance directions. Only when the original target bubble region also produces morphological changes that meet the requirements of response direction, response time difference and linkage response intensity during the reverse disturbance is it further confirmed that the two are connected through the gas channel under the tag, thereby reducing the possibility of misjudgment caused by unidirectional non-gas migration responses such as local elastic transmission of the tag and overall mechanical deformation caused by pressure contact.

[0025] Secondly, this application provides a label bubble recognition system for label caps, used to execute any of the above-described label bubble recognition methods. The system includes an image acquisition device, a pressure application component, and a processing device. The image acquisition device is used to acquire a label image of the label cap and acquire a response label image during and after the pressure application component applies a local recoverable pressure disturbance to the target bubble area. The pressure application component is used to apply a local recoverable pressure disturbance to the target bubble area in the candidate bubble area. The processing device is used to identify multiple mutually separated candidate bubble areas from the label image and determine their spatial proximity based on their spatial location. It also determines the morphological changes of the target bubble area and the remaining candidate bubble areas separated from it based on the response label image, determines the connected bubble group connected through the gas channel below the label based on the linkage response relationship generated by the pressure disturbance, and generates a label bubble recognition result and determines the bonding quality status of the label cap based on the overall coverage of the connected bubble group.

[0026] Compared with related technologies, this application has the following advantages: First, separate candidate bubble regions are identified from the label image and their spatial proximity is determined. Then, a local recoverable pressure perturbation is applied to the target bubble region, and the morphological changes of the target bubble region and other spatially adjacent candidate bubble regions during the pressure perturbation and after the pressure is released are simultaneously acquired. This allows the gas connectivity state under the label, which cannot be directly observed from a single two-dimensional image, to be transformed into a dynamic morphological response with temporal correlation between candidate bubble regions. Furthermore, the gas connectivity relationship between visually separated regions is determined based on the linkage response relationship between the target bubble region and other candidate bubble regions with pressure perturbation. Candidate bubble regions with direct or indirect connectivity are treated as connected bubble groups for overall coverage evaluation. This allows the label bubble recognition object to be further mapped from a simple two-dimensional visible connected domain to the actual gas connectivity structure, reducing the possibility of missing overall bubble defects due to the separate evaluation of multiple visually separated small bubbles, and improving the consistency between the bonding quality judgment and the actual bonding state of the label. Attached Figure Description

[0027] Figure 1 A schematic diagram of a label bubble and a gas channel below the label provided in an embodiment of this application; Figure 2 A flowchart illustrating the overall process of a label bubble recognition method provided in this application embodiment; Figure 3 A flowchart for determining the bonding quality status is provided in an embodiment of this application; Figure 4 This is a schematic diagram of a label bubble recognition system provided in an embodiment of this application. Detailed Implementation

[0028] To make the objectives, technical solutions, and technical effects of the embodiments of this application clearer, the embodiments of this application will be further described below with reference to the accompanying drawings. The described embodiments are used to explain this application and are not intended to limit the scope of protection of this application.

[0029] In this embodiment, the label cap refers to a product with a label affixed to its surface. The label image refers to an image that at least covers the area of ​​the label to be detected. The candidate bubble area refers to an area identified from the label image as potentially having a separation space between the label and the cap, based on the brightness, edges, texture, or other image features that can characterize a local bulge.

[0030] In this embodiment, mutual separation or spatial separation describes the two-dimensional connectivity of candidate bubble regions in the label image, meaning that two candidate bubble regions belong to different two-dimensional connected regions in the bubble region segmentation result; spatial proximity describes the positional distance relationship between two separated candidate bubble regions on the label surface. Therefore, two candidate bubble regions can be separated in the two-dimensional image, while having a spatial proximity relationship because their actual positional distance is no greater than a preset spatial proximity threshold.

[0031] The locally recoverable pressure disturbance in this application embodiment refers to a pressure disturbance that, when applied to the label surface corresponding to the target bubble area, causes a morphological change in the target bubble area exceeding the image measurement noise, and after the pressure is released, the target bubble area can recover to a preset recovery range.

[0032] The bonding quality status in this application embodiment is determined based on label bubble defects, and is used to characterize whether the label cap meets the corresponding quality requirements under the label bubble detection item; label misalignment, damage, wrinkles or other non-bubble defects can be judged independently through the corresponding detection process.

[0033] like Figure 1As shown, the two candidate bubble regions may not have continuous abnormal contours or textures in the label image, thus appearing as separate two-dimensional images. However, between the label and the cover, the two candidate bubble regions may still be connected by a gas channel that is too low or too narrow to meet the candidate bubble recognition criteria for the corresponding label region. When one candidate bubble region is subjected to pressure, the gas inside it can transfer to the other candidate bubble region through the gas channel, thereby forming a corresponding morphological linkage.

[0034] In one application scenario, a label cap to be inspected is placed at an inspection station. The imaging field of view of the image acquisition device at least covers the label area to be inspected on the label cap. A pressure application component is used to apply local recoverable pressure perturbation to the target bubble area in the label area to be inspected. Both the image acquisition device and the pressure application component are connected to a processing device. The processing device acquires the label image and the response label image formed by the image acquisition device, and controls or records the pressure perturbation process of the pressure application component. Based on the label image, the response label image, and the pressure perturbation process, it generates the label bubble recognition result and the bonding quality status.

[0035] Specifically, the image acquisition device can be an industrial camera, an area scan camera, or other imaging device capable of obtaining images of the surface of the label to be inspected.

[0036] Specifically, the pressure application component may include a drive element, a pressure detection element, and a pressure contact end. To ensure that the response tag image of the target bubble region can still be acquired during pressure application, the pressure contact end may be transparent, allowing imaging light to pass through at least within the imaging optical path range corresponding to the image acquisition device. The imaging field of view of the image acquisition device covers the target bubble region acted upon by the transparent pressure contact end and other candidate bubble regions spatially adjacent to it, enabling the pressure application component to acquire the corresponding response tag image while maintaining pressure.

[0037] In another embodiment, the pressure application component and the image acquisition device can be staggered. The pressure application component applies pressure along the normal direction of the label surface or in a direction inclined relative to the label surface. The image acquisition device acquires the response label image from an imaging direction not obstructed by the pressure application component, and performs geometric correction and position registration on the response label image in subsequent steps. The improvement focus of this application is not on the specific structure of the pressure contact end and the image acquisition device, but on ensuring that the target bubble area remains imageable during pressure disturbance, thereby obtaining the dynamic image data required for subsequent linkage response judgment.

[0038] like Figure 2 As shown, a method for identifying bubble marks on a label includes steps S101 to S105.

[0039] Step S101: Obtain the label image of the labeling cap, identify multiple mutually separated candidate bubble regions, and determine the spatial proximity relationship based on the spatial position between each candidate bubble region.

[0040] The detailed implementation steps of step S101 include: Step S1011: Acquire the label image of the labeling cap. Specifically, the labeling cap to be inspected is transported or set to a predetermined inspection station, and after the labeling cap reaches the inspection position, the image acquisition device is triggered to acquire at least one label image covering the label area to be inspected. For continuous production lines, image acquisition can be triggered based on the arrival signal provided by the position sensor, the equipment controller, or the production cycle signal; for offline inspection, image acquisition can be triggered after the labeling cap is fixed in the inspection position.

[0041] Specifically, when the image acquisition device acquires a tag image, it simultaneously records an image timestamp, which is used to establish an acquisition timing relationship with subsequent response tag images.

[0042] Before calculating the bubble region size, the image acquisition device is calibrated. If the label area to be detected can be treated as a plane, the horizontal scaling factor s of the image can be determined using a calibration piece with known actual dimensions or a structure on the label cover with known actual dimensions. x And the vertical scaling factor s of the image y .

[0043] For the pixel coordinates (u,v) in the label image, the corresponding actual planar coordinates are determined according to the following formula:

[0044]

[0045] Where u represents the pixel coordinates in the horizontal direction of the image; y represents the vertical pixel coordinates of the image; x represents the actual horizontal position coordinates of the corresponding pixel on the label surface; y represents the actual vertical position coordinates of the corresponding pixel on the label surface; s x This represents the actual length corresponding to a unit pixel in the horizontal direction of the image; s y This represents the actual length corresponding to a unit pixel in the vertical direction of the image.

[0046] When the candidate bubble region contains N i When the area is 1 pixel, its actual region area is determined by the following formula:

[0047] Among them, A i N represents the actual area of ​​the i-th candidate bubble region; iIndicates the first The number of pixels contained in each candidate bubble region; Indicates the candidate bubble region number.

[0048] When the label to be detected is located on the surface of a cover with non-negligible curvature, a positional mapping relationship between the image pixel position and the actual position on the label surface can be established by calibrating samples. Based on the positional mapping relationship, the pixel coordinates of each candidate bubble region can be converted into the actual position.

[0049] Therefore, step S1011 obtains the label image of the labeling cap, and the label image and its size calibration relationship are used by step S1012 to identify candidate bubble regions and by step S1013 to calculate the actual distance of candidate bubble regions.

[0050] Step S1012: Identify multiple mutually separated candidate bubble regions; specifically, bubble region segmentation can be performed based on at least one of local brightness differences, abnormal edges, and abnormal textures in the label image.

[0051] In one implementation, a normal grayscale reference corresponding to the position of the label image to be detected can be pre-formed using multiple normal fitting calibration samples. For pixels or local image blocks in the label image to be detected, the normal grayscale reference value corresponding to the same image position is read, and the absolute difference between the grayscale value to be detected and the normal grayscale reference value is calculated. When the absolute difference is not less than the bubble recognition threshold, the corresponding pixel is marked as an abnormal pixel, connected component marking is performed on adjacent abnormal pixels, and connected components with an actual area not less than the minimum detection area are determined as candidate bubble regions.

[0052] Let the mean of the grayscale difference corresponding to the normal bonding calibration sample be . The standard deviation is The bubble recognition threshold is... One way it is formed is as follows:

[0053] in, Indicates the bubble recognition threshold; This represents the mean grayscale difference of normally fitted calibration samples; This represents the standard deviation of the grayscale difference between normally fitted calibration samples.

[0054] The above-mentioned double standard deviation is used to provide an example of an executable threshold formation, and the initial bubble recognition threshold can also be verified using confirmed normal fit samples and bubble samples.

[0055] In another implementation, abnormal edges in the label image can be detected. When an abnormal edge forms a closed boundary, the region inside the closed boundary is designated as the region to be judged. When an abnormal edge has unconnected endpoints, the actual distance between two endpoints to be connected is calculated. If the actual distance is not greater than a preset boundary connection distance, and connecting the two endpoints forms a closed region, the connected closed region is designated as the region to be judged. Then, based on the image differences between the region to be judged and the normally fitted region, it is determined whether the corresponding region to be judged is identified as a candidate bubble region.

[0056] Specifically, for each region to be judged formed by abnormal edges, the grayscale values ​​of each pixel within the region to be judged are read and the average grayscale value is calculated. At the same time, normally fitted regions at the same label position as the region to be judged are read from the normal fitting calibration samples and their average grayscale values ​​are calculated. The absolute difference between the two average grayscale values ​​is determined as the region grayscale difference. The grayscale difference threshold of the region The value can be determined based on the statistical results of the grayscale difference in the corresponding areas of the same label position among multiple normally fitted calibration samples, for example, by calculating the mean of the grayscale difference set. and standard deviation and in accordance with Determine the initial grayscale difference threshold for the region, and then verify it using confirmed properly fitted samples and confirmed bubble samples; when the region to be judged meets the following conditions... If the actual area is not less than the minimum detection area, the area to be judged is determined as a candidate bubble area; otherwise, it is excluded. The minimum detection area can be determined based on the actual spatial resolution of the image acquisition device and the minimum bubble size to be detected in the product. When it is impossible to obtain normal bonding calibration samples with the same label position, the boundary of the area to be judged can be extended outward by a preset reference width to form an outer reference area. After removing other abnormal pixels, the remaining pixels are used as local normal bonding reference pixels, and the reference gray value of the normal bonding area is determined based on its average gray value. Thus, the candidate bubble area used for subsequent spatial proximity determination is obtained.

[0057] The preset boundary connection distance can be determined based on the statistical results of the natural break length of the edge in the normal label image, or based on the minimum boundary length that the image acquisition device can reliably identify.

[0058] For the There are 1 candidate bubble regions, and their actual location regions are represented as follows: and determine its regional boundaries. and the baseline area before pressure application .

[0059] in, Indicates the first One candidate bubble region; Indicates the first The region boundaries of each candidate bubble region; Indicates the first The actual baseline area of ​​each candidate bubble region before pressure is applied, and since the corresponding object has already been identified as a candidate bubble region, therefore:

[0060] Therefore, step S1012 identifies multiple mutually separated candidate bubble regions, and obtains the regional location, regional boundary and reference area of ​​each candidate bubble region, which are used by step S1013 to determine the spatial proximity relationship and by step S102 to determine the target bubble region.

[0061] Step S1013: Determine the spatial proximity relationship based on the spatial location between each candidate bubble region. Specifically, for any two candidate bubble regions... and Calculate the minimum boundary distance between the two regions based on the boundary positions of the regions after dimensional calibration:

[0062] in, Indicates the first The candidate bubble region and the first Minimum actual boundary distance between candidate bubble regions; Indicates the first Location points on the boundary of each candidate bubble region; Indicates the first Location points on the boundary of each candidate bubble region; Indicates Euclidean distance; .

[0063] Will Proximity threshold of preset space Compare.

[0064] when: At that time, in the The candidate bubble region and the first Establish spatial proximity relationships between candidate bubble regions.

[0065] when: When that happens, the two corresponding candidate bubble regions are not identified as spatially adjacent bubble region pairs.

[0066] Preset spatial proximity threshold This can be determined by identifying pairs of calibration bubble regions with confirmed gas connectivity. The minimum boundary distances of multiple pairs of confirmed gas-connected calibration bubble regions are statistically analyzed, and an initial spatial proximity threshold is determined based on a preset quantile value of the corresponding distance distribution. For example, the 95th percentile value of the minimum boundary distance of confirmed gas-connected samples can be used as the initial spatial proximity threshold, and then the initial spatial proximity threshold is verified using confirmed non-connected samples.

[0067] In step S101, candidate bubble regions to be verified are identified using the two-dimensional image separation results, and spatial proximity relationships are further formed based on the actual distance, so that subsequent pressure disturbance analysis focuses on candidate bubble regions with potential hidden gas connectivity.

[0068] For example, when processing a label cap to be inspected in step S101, the image acquisition device acquires a label image covering the entire label area. After size calibration, the actual length of each pixel in the horizontal and vertical directions of the image is determined to be 0.02 mm. After performing bubble region segmentation on the label image, three candidate bubble regions A, B, and C, which are separated from each other in the two-dimensional image, are obtained. The reference area before pressure application for candidate bubble region A is 1.80 mm², for candidate bubble region B it is 1.70 mm², and for candidate bubble region C it is 1.20 mm². The minimum actual boundary distance is calculated based on the region boundaries of each candidate bubble region. The minimum boundary distance between candidate bubble region A and candidate bubble region B is 2.0 mm, between candidate bubble region A and candidate bubble region C it is 5.4 mm, and between candidate bubble region B and candidate bubble region C it is 4.9 mm. If the spatial proximity threshold predetermined based on the calibration samples is 3.0 mm, then since the minimum boundary distance between candidate bubble region A and candidate bubble region B is no greater than 3.0 mm, a spatial proximity relationship is established between candidate bubble region A and candidate bubble region B; the minimum boundary distance between candidate bubble region C and candidate bubble regions A and B are both greater than 3.0 mm, therefore, no corresponding spatial proximity relationship is established at this time. Thus, step S101 obtains candidate bubble regions A, B, and C and their spatial proximity relationships, and selects candidate bubble regions A and B, which have spatial proximity relationships, as the objects for priority pressure perturbation verification in step S102.

[0069] Step S102: Apply a local recoverable pressure perturbation to the target bubble region within the candidate bubble region, and acquire response label images during the pressure perturbation and after the pressure is released; the detailed implementation steps of step S102 include: Step S1021: Determine the target bubble region. Specifically, from the multiple candidate bubble regions obtained in step S101, select candidate bubble regions that have at least one spatially adjacent candidate bubble region and have not yet completed the gas connectivity determination, and determine one candidate bubble region as the target bubble region from the selection results.

[0070] Specifically, the target bubble regions can be determined sequentially according to their candidate bubble region numbers, or the detection order can be determined by sorting the candidate bubble regions according to their actual areas. The sorting method for target bubble regions is only used to determine the execution order of pressure disturbances and does not participate in the determination of gas connectivity itself.

[0071] After determining the target bubble region, candidate bubble regions that have a spatial proximity relationship with the target bubble region are read from the spatial proximity relationship obtained in step S101, and are used as the objects to be observed in step S103.

[0072] Therefore, step S1021 determines the target bubble region and its corresponding spatially adjacent candidate bubble region, so that steps S1022 and S1023 can perform pressure perturbation.

[0073] Step S1022: Determine the recoverable pressure range and preset disturbance duration of the local recoverable pressure disturbance. Specifically, the pressure detection element of the pressure application component collects the contact force exerted by the pressure contact end on the label surface corresponding to the target bubble area. Based on the effective contact area between the pressure contact end and the label surface, the contact force is converted into a disturbance pressure value.

[0074] in, This indicates the disturbance pressure value acting on the label surface corresponding to the target bubble area; This indicates the contact force collected by the pressure sensor; This indicates the effective contact area between the pressure contact end and the label surface, and:

[0075] When the pressure sensor can directly output the pressure value, the corresponding pressure value can be directly read as the disturbance pressure value.

[0076] Before formal testing, the recoverable pressure range is determined using calibration samples. Starting from a preset initial pressure, the perturbation pressure value is increased step by step according to a predetermined pressure increment. After each pressure level is applied, the pressure is released, and the recovered image of the corresponding calibration bubble region is obtained.

[0077] The rate of change of residual area after pressure relief is determined by the following formula:

[0078] in, This represents the rate of change of the residual area after the pressure is released; This represents the reference area of ​​the corresponding calibrated bubble region before pressure is applied, and ; This indicates the area of ​​the region after the pressure has been relieved and a preset recovery period has elapsed.

[0079] To determine the preset recovery threshold, multiple reference images were continuously acquired for the same candidate bubble region without applying pressure. The area corresponding to one of the reference images was used as the reference area. , will the The area corresponding to the repeated acquisition of images is denoted as The fluctuation rate of the repeated imaging area is calculated according to the following formula:

[0080] in, Indicates the first Area fluctuation rate corresponding to repeated image acquisition; Indicates the first The area of ​​candidate bubble regions in repeatedly acquired images; This indicates the image number that was captured repeatedly.

[0081] Statistical analysis of the area fluctuation of multiple repeating images The mean area volatility was obtained. and standard deviation The preset recovery threshold is formed according to the following formula:

[0082] in, This indicates the preset recovery threshold.

[0083] When the morphological change during the pressure application phase corresponding to a perturbation pressure value exceeds the image measurement noise range, and the following conditions are met after the pressure is released: When the disturbance pressure value is determined, it is defined as the recoverable pressure value. The recoverable pressure range consists of multiple consecutive recoverable pressure values.

[0084] The preset perturbation duration is further determined. For continuously acquired response label images, the change in the area of ​​the target bubble region corresponding to two adjacent acquisition times is determined as follows:

[0085] in, The table represents the first The area change of adjacent frames corresponding to each acquisition time; Indicates the target bubble region in the th... The area of ​​the region at each data collection time; The table represents the area of ​​the target bubble region at the previous data acquisition time; This indicates the sequence number of the response image acquisition.

[0086] When continuous All data collection times satisfy the following:

[0087] When this occurs, the corresponding stage is defined as the morphology response preservation stage.

[0088] in, This represents the preset area stabilization threshold, which can be determined based on the statistical results of area changes in adjacent frames obtained from repeated imaging under no pressure. The preset number of consecutive judgments is not less than 2.

[0089] The preset perturbation duration can be determined based on the statistical results of the time taken from the start of pressure application to the morphological response maintenance phase of multiple calibration samples. For example, the 95th percentile of the corresponding time distribution can be used as the initial value of the preset perturbation duration.

[0090] Therefore, step S1022 determines the recoverable pressure range and preset disturbance duration of the local recoverable pressure disturbance, so that step S1023 can actually execute the pressure disturbance.

[0091] Step S1023: Apply a local recoverable pressure disturbance to the target bubble area. Specifically, based on the target bubble area position determined in step S1021, control the pressure application component to move the pressure contact end to the label surface corresponding to the target bubble area, and apply a disturbance pressure value within the recoverable pressure range determined in step S1022.

[0092] The processing device reads the real-time pressure detection value of the pressure sensor. When the pressure detection value enters the target disturbance pressure range, the preset disturbance duration is maintained.

[0093] When the pressure detection value is lower than the lower limit of the target disturbance pressure range, the pressure application component continues to adjust the pressure.

[0094] When the pressure detection value exceeds the upper limit of the recoverable pressure range, the current pressure disturbance is marked as an invalid disturbance, the pressure increase is immediately stopped, and the current pressure effect is released; the response label image collected during the current invalid disturbance period does not proceed to step S103. After the target bubble region meets the preset recovery conditions again, the pressure disturbance of the current target bubble region is re-executed using the disturbance pressure value within the recoverable pressure range.

[0095] Therefore, step S1023 applies a local pressure disturbance to the target bubble region that meets the recoverable pressure requirement, and the data corresponding to the effective pressure disturbance is used by step S1024 to form a response label image.

[0096] Step S1024: Acquire response tag images during the pressure disturbance and after the pressure is released. Specifically, acquire at least one tag image as a reference for response analysis before the pressure is applied; acquire response tag images of the pressure application phase continuously at a preset image acquisition interval from the moment the pressure is applied; and continue to acquire response tag images of the recovery phase at the preset image acquisition interval after the pressure is released, until the preset recovery period is reached.

[0097] Specifically, during a single pressure disturbance process corresponding to the same target bubble region, the imaging magnification, exposure parameters, and image gain of the image acquisition device are kept constant to reduce the impact of changes in imaging conditions on the boundary recognition results of the candidate bubble region.

[0098] The image acquisition device or processing device records a corresponding timestamp for each response label image. The timestamp is used in step S104 to determine the response start time and response time difference of the target bubble region and the other candidate bubble regions.

[0099] The interval for acquiring the response label image is denoted as . To ensure that the sequence of responses during the stress and recovery phases can be characterized by at least two sampling intervals, the following can be made:

[0100] in, Indicates the maximum permissible time difference for stress response; Indicates the maximum permissible recovery response time difference; This represents the smaller of the two values.

[0101] The and All parameters are pre-formed during the calibration phase and used as response time parameters during the formal testing process.

[0102] The response tag images before, during, and after pressure application are sorted by timestamp to form a perturbation response image sequence corresponding to the current target bubble region.

[0103] Therefore, step S1024 acquires response tag images during the pressure disturbance and after the pressure is released, forming a disturbance response image sequence containing timing information of the pressure application phase and the recovery phase, which is used by step S103 to determine the corresponding morphological changes of the target bubble region and the remaining candidate bubble regions.

[0104] For example, following the spatial proximity relationship obtained in step S101, the processing device first determines candidate bubble region A as the target bubble region, and determines candidate bubble region B, which has a spatial proximity relationship with candidate bubble region A, as the subsequent synchronous observation object. Assuming the recoverable pressure range pre-calibrated for the current label material and pressure contact end is 12kPa to 18kPa, and the preset disturbance duration is 0.8s, the processing device controls the pressure application component to move the pressure contact end to the label surface corresponding to candidate bubble region A, and gradually increases the pressure; when the pressure detection component detects that the disturbance pressure value reaches 15kPa, it maintains this disturbance pressure for 0.8s. Assuming the maximum allowable pressure response time difference is 0.20s and the maximum allowable recovery response time difference is 0.25s, the acquisition interval of the response label image is set to 0.05s, ensuring that 0.05s is not greater than half the smaller of 0.20s and 0.25s. Before applying pressure, a baseline label image is acquired. Starting from the moment pressure is applied, a pressure-stage response label image is acquired every 0.05 seconds. After the pressure is maintained for 0.8 seconds, the pressure is released, and recovery-stage response label images are acquired at 0.05-second intervals until the end of the preset recovery period. Each response label image is recorded with a corresponding timestamp. Assuming that candidate bubble region A recovers to the preset recovery range after the pressure is released, the pressure disturbance is determined to be a valid pressure disturbance. The images before pressure application, the pressure-stage response label images, the recovery-stage response label images after pressure release, and the corresponding timestamps are arranged in the acquisition order to form a disturbance response image sequence, which is used in step S103 to determine the morphological changes of candidate bubble region A and candidate bubble region B.

[0105] Step S103: Determine the morphological changes of the target bubble region based on the response label image, and filter out other candidate bubble regions spatially separated from the target bubble region based on the spatial proximity relationship, and determine the corresponding morphological changes of the filtered other candidate bubble regions. The detailed implementation steps of step S103 include: Step S1031: Perform position registration on the response label image. Specifically, using the label image obtained before pressure is applied as the reference image, determine registration feature points from the corner points of the label printing pattern, the feature points of the label outer contour, or the structural feature points on the cover that do not belong to the pressure application area, and determine the position corresponding to the registration feature points in each response label image.

[0106] When image translation and rotation mainly occur at the detection station, the two-dimensional rigid transformation relationship between the response label image and the reference image can be calculated based on at least two non-overlapping registration feature points, and the response label image can be transformed to the reference image coordinate system based on the two-dimensional rigid transformation relationship.

[0107] When the imaging direction is tilted relative to the label surface, the geometric transformation relationship can be determined based on multiple registration feature points. After performing geometric correction on the response label image, position registration can be performed.

[0108] Step S1032: Determine the morphological changes of the target bubble region. Specifically, after completing the position registration, expand the target bubble region's boundary outward by a preset search distance to form a target tracking region. Within the target tracking region corresponding to each response label image, redetermine the abnormal region according to the bubble region identification rules consistent with step S1012.

[0109] When there are multiple abnormal regions within the same target tracking area, calculate the overlap area ratio between each abnormal region and the target bubble region at the previous acquisition time, and determine the abnormal region with the largest overlap area ratio as the target bubble region corresponding to the current acquisition time.

[0110] The preset search distance can be determined based on the statistical results of the maximum bubble boundary displacement caused by recoverable pressure disturbance during the calibration process. For example, the 95th percentile value of the maximum bubble boundary displacement can be used as the initial search distance.

[0111] After obtaining the target bubble region frame by frame, determine the actual area of ​​the region corresponding to each acquisition time. .

[0112] in, Indicates the first Each candidate bubble region at the time of data acquisition The actual area of ​​the region; This indicates the timestamp of the image capture for the corresponding response label.

[0113] The minimum area of ​​the target bubble region detected during the pressure application phase is denoted as... The contraction response of the target bubble region is determined by the following formula:

[0114] in, This represents the contraction response of the target bubble region; This represents the baseline area of ​​the target bubble region before pressure is applied; This represents the minimum area of ​​the target bubble region during the pressure application phase.

[0115] The recovery area corresponding to the target bubble region within the preset recovery period after pressure relief is denoted as [missing information]. The recovery response of the target bubble region is determined according to the following formula:

[0116] in, This represents the recovery response of the target bubble region; This indicates the recovery area of ​​the target bubble region after the pressure is released.

[0117] In one embodiment for reducing the impact of single-frame segmentation fluctuations, multiple target bubble region areas can be acquired within the shape response holding phase determined in step S1022, and the average or median of the multiple areas can be used as the representative area of ​​the pressure application phase, replacing the representative area of ​​the pressure application phase. Calculate the contraction response.

[0118] When equivalent dimensions are needed to help describe morphological changes, the diameter of the equivalent circle can be determined based on the area of ​​the region.

[0119] in, Indicates the first The candidate bubble regions at time [time] The equivalent circle diameter; Pi is used to represent the mathematical constant pi. The equivalent circle diameter is used to uniformly characterize the equivalent size of bubble regions with different contours, and does not mean that the actual contour of the corresponding candidate bubble region must be circular.

[0120] It can also calculate the boundary displacement based on the position changes of the boundary points of the candidate bubble region at each acquisition time, and use the area change and boundary displacement together to describe the morphological change.

[0121] Therefore, step S1032 determines the morphological changes of the target bubble region during the pressure application and pressure release phases, so that step S104 can determine the contraction and recovery responses of the target bubble region.

[0122] Step S1033: Based on the spatial proximity relationship, filter out the remaining candidate bubble regions that are spatially separated from the target bubble region; specifically, read the spatial proximity relationship established in step S101, filter out candidate bubble regions that have a spatial proximity relationship with the current target bubble region and belong to different two-dimensional connected regions in the bubble region segmentation result of the label image before pressure application, and determine the filtered candidate bubble regions as the remaining candidate bubble regions. Thus, step S1033 filters out the remaining candidate bubble regions that need to be linked with the current target bubble region for analysis based on the spatial proximity relationship, for step S1034 to determine the corresponding morphological changes.

[0123] Step S1034: Determine the corresponding morphological changes of the remaining candidate bubble regions. Specifically, for the first... For the remaining candidate bubble regions, establish corresponding tracking regions based on their pre-pressure region positions, and continuously determine the corresponding regions in each response label image using the same frame-by-frame tracking method as in step S1032.

[0124] The first The baseline area of ​​the remaining candidate bubble regions before pressurization is denoted as . The target bubble region will be under pressure during the first The maximum area reached by the remaining candidate bubble regions is denoted as . The recovery area after pressure relief is recorded as . .

[0125] Its extended response is determined according to the following formula:

[0126] in, Indicates the first indivual The expansion response of the remaining candidate bubble regions during the compression of the target bubble region; This indicates the maximum area reached by the other candidate bubble regions during the pressure application phase.

[0127] Its reverse recovery response is determined according to the following formula:

[0128] in, Indicates the first The reverse recovery response of the remaining candidate bubble regions after pressure relief; This represents the recovery area of ​​the remaining candidate bubble regions after pressure relief.

[0129] In one embodiment for reducing the impact of segmentation fluctuations in a single frame image, the areas of multiple remaining candidate bubble regions can be obtained within the morphology response retention interval during the pressure application phase, and the average or median area can be used as the representative area during the pressure application phase, replacing... Calculate the extended response.

[0130] To distinguish between actual morphological changes and image measurement fluctuations, multiple baseline images of the same candidate bubble region can be acquired consecutively before pressure is applied, and the standard deviation of the measurement fluctuation of the corresponding region area can be statistically analyzed. And determine the area response noise threshold:

[0131] in, Indicates the area response noise threshold; This represents the standard deviation of the area measurement fluctuation of the corresponding candidate bubble region under unpressurized conditions; This represents a dimensionless confidence coefficient, which can be 2 or 3, and is calibrated according to the permissible false trigger level for repeated detection.

[0132] When the absolute value of the area change of the corresponding candidate bubble region at a certain stage is greater than When the morphological change is determined, the corresponding morphological change is identified as the effective morphological response.

[0133] In the presence of hidden gas channels, the target bubble region decreases in area after being pressurized, while the other candidate bubble regions increase in area after the corresponding response time difference; after the pressure is released, the target bubble region recovers in area, while the other candidate bubble regions show a recovery change in the opposite direction to the pressurization phase.

[0134] Therefore, step S1034 determines the corresponding morphological changes of the remaining candidate bubble regions, and the morphological changes of the target bubble region and the corresponding morphological changes of the remaining candidate bubble regions are used together as inputs for step S104 to determine the linkage response relationship.

[0135] For example, based on the disturbance response image sequence formed in step S102, using the pre-pressure label image as the reference image, position registration is performed on each response label image, and bubble regions are extracted frame by frame within the tracking areas corresponding to candidate bubble region A and candidate bubble region B, respectively. Assuming the reference area of ​​candidate bubble region A before pressure is 1.80 mm², its area gradually decreases during the pressure application stage, and a representative area of ​​1.35 mm² is obtained during the morphology response maintenance stage, then the contraction response amount of candidate bubble region A is (1.80−1.35) / 1.80=0.25; after the pressure is released, candidate bubble region A recovers to 1.76 mm², then its recovery response amount is (1.76−1.35) / 1.80≈0.228. Meanwhile, the baseline area of ​​candidate bubble region B before pressure application is 1.70 mm². During the pressure application period of candidate bubble region A, its area increases from 1.70 mm² to a representative area of ​​2.04 mm² during the pressure application phase. Therefore, the expansion response of candidate bubble region B is (2.04 − 1.70) / 1.70 = 0.20. After the pressure is released, the area of ​​candidate bubble region B recovers from 2.04 mm² to 1.73 mm², and its reverse recovery response is (2.04 − 1.73) / 1.70 ≈ 0.182. If the area change rate corresponding to the area response noise threshold obtained by repeated imaging before pressure application is less than the above effective change amounts, then the decrease in area of ​​candidate bubble region A is determined as an effective contraction response, the increase in area of ​​candidate bubble region B is determined as an effective expansion response, and the corresponding changes after pressure release are determined as the effective recovery response and the effective reverse recovery response, respectively.

[0136] Step S104: Based on the linkage response relationship between the target bubble region and the other candidate bubble regions caused by the pressure disturbance, determine the visually separated but connected bubble group through the gas channel below the label; the detailed implementation steps of step S104 include: Step S1041: Determine the response time difference between the target bubble region and the other candidate bubble regions. Specifically, during the pressure application phase, the moment when the area change of the target bubble region first exceeds its area response noise threshold, and the direction of change is area reduction, is determined as the starting moment of the target contraction response. ; will the first The moment when the area change of the remaining candidate bubble regions first exceeds the corresponding area response noise threshold, and the direction of change is increasing area, is determined as the starting moment of the neighboring expansion response. .

[0137] The time difference for pressure response is determined by the following formula:

[0138] in, This indicates that the target bubble region has begun to shrink to the [number]th [unit]. The time difference of pressure response between the start of expansion of the remaining candidate bubble regions; Indicates the start time of the target contraction response; This indicates the start time of the neighboring expansion response.

[0139] During the pressure relief phase, the moment when the area change of the target bubble region first exceeds the corresponding area response noise threshold, and the direction of change is an increase in area, is defined as the start moment of the target recovery response. ; will the first The moment when the area change of the remaining candidate bubble regions first exceeds the corresponding area response noise threshold, and the change direction is a decrease in area, is determined as the starting moment of the nearest inverse recovery response. .

[0140] The recovery response time difference is determined by the following formula:

[0141] in, This indicates that after the pressure is released, the target bubble area begins to recover to the previous state. The recovery response time difference between the start of reverse recovery for each of the remaining candidate bubble regions.

[0142] Only when:

[0143] and:

[0144] At that time, determine the corresponding bubble region pair that meets the response time condition.

[0145] in, Indicates the maximum permissible time difference for stress response; This indicates the maximum permissible recovery response time difference.

[0146] when When this occurs, it indicates that the expansion response of the remaining candidate bubble regions precedes the effective contraction response of the target bubble region, which does not conform to the response timing of subsequent gas migration triggered by pressure in the target region; when When this occurs, it indicates that the corresponding extended response exceeds the allowable response time range. The recovery phase is handled according to the same rules.

[0147] Step S1042: Determine the linkage response intensity. Specifically, calculate the linkage response intensity only for candidate bubble regions that meet the response time condition in step S1041.

[0148] The linkage response intensity is used to evaluate whether the two candidate bubble regions simultaneously exhibit morphological responses corresponding to gas redistribution during the pressurization and recovery phases, and is not used to calculate the actual transfer volume or actual gas flow rate of the gas below the label.

[0149] In one implementation, the intensity of the linkage response is determined according to the response amplitudes during the pressure application phase and the pressure release phase, using the following formula:

[0150] in, Indicates the target bubble region and the first The intensity of the linkage response between the remaining candidate bubble regions; This represents the contraction response of the target bubble region; Indicates the first The expansion response of the remaining candidate bubble regions; This represents the recovery response of the target bubble region; Indicates the first The reverse recovery response of the remaining candidate bubble regions; This represents the effective linkage response amplitude threshold during the pressure application phase, and:

[0151] This represents the effective linkage response amplitude threshold during the recovery phase, and:

[0152] Indicates the weighting of the response evaluation during the pressure application phase; This indicates the weight of the response evaluation during the recovery phase.

[0153] The weights satisfy:

[0154] and:

[0155] Without setting different evaluation priorities for the two stages, we can let:

[0156] Since each component is truncated and normalized using a corresponding response amplitude threshold, therefore:

[0157] Specifically, the effective linkage response amplitude threshold during the pressure application phase. This can be formed using the statistical results of the non-linkage area change rate generated by confirmed gas-disconnected samples during pressure disturbances. For example, the mean of the corresponding non-linkage area change rate plus twice the standard deviation can be used as the initial value; the effective linkage response amplitude threshold during the recovery phase. The same method can be used to form the statistical results of the non-linkage area change rate during the pressure relief phase.

[0158] Therefore, step S1042 determines the intensity of the linkage response, and step S1043 determines whether there is a gas connection relationship between the corresponding candidate bubble regions.

[0159] Step S1043: Determine the gas connectivity between candidate bubble regions based on the aforementioned linkage response relationship; specifically, when the target bubble region forms an effective contraction response during the pressure application phase, the... The remaining candidate bubble regions form an effective expansion response within the maximum permissible pressure response time difference, and the target bubble region forms an effective recovery response after the pressure is released. When the remaining candidate bubble regions form an effective reverse recovery response within the maximum allowable recovery response time difference, the corresponding two candidate bubble regions will be determined as a linkage response with two-stage directional correspondence.

[0160] Furthermore, the intensity of the linkage response will be increased. With preset linkage response strength threshold Compare them.

[0161] when:

[0162] Furthermore, when the corresponding candidate bubble region simultaneously satisfies the response direction condition of the pressure application phase, the response direction condition of the recovery phase, the pressure application response time condition, and the recovery response time condition, the target bubble region is determined to be related to the first... The remaining candidate bubble regions have gas connectivity.

[0163] Specifically, a preset linkage response strength threshold is defined. The threshold is determined based on the intensity distribution of the linkage response of confirmed gas-connected and confirmed gas-disconnected samples. When the intensity ranges of the linkage response of the two types of calibration samples do not overlap, a threshold can be selected between the maximum linkage response intensity of the gas-disconnected sample and the minimum linkage response intensity of the gas-connected sample. When the two distributions partially overlap, multiple candidate thresholds can be used to perform calibration sample judgment, and the candidate threshold with fewer incorrect judgments can be selected as the preset linkage response intensity threshold. .

[0164] Therefore, step S1043 determines the gas connectivity relationship between candidate bubble regions based on the linkage response relationship. The gas connectivity relationship is used by step S1044 for optional reverse verification or by step S1045 to directly construct a connected bubble group.

[0165] Step S1044: Perform reverse pressure perturbation verification on the gas connectivity relationship. Specifically, in order to reduce misjudgment caused by non-gas migration response in adjacent bubble areas due to local elastic deformation of the label, when step S1043 initially determines that there is a gas connectivity relationship between the target bubble area and any other candidate bubble area, the current pressure can be released and the corresponding bubble area can be waited to recover to the preset recovery range. Then, the other candidate bubble areas are treated as new target bubble areas and local recoverable pressure perturbation is applied.

[0166] Following the same processing method as steps S103 and S1041 to S1043, determine whether the original target bubble region produces an expansion response during the new target bubble region under pressure, and whether it produces a reverse recovery response after the pressure is released, and calculate the corresponding response time difference and linkage response intensity.

[0167] When the response time difference corresponding to the reverse pressure disturbance is within the corresponding allowable time range, and the linkage response intensity is not less than the preset linkage response intensity threshold, the two candidate bubble regions are confirmed to be connected through the gas channel below the label.

[0168] If only the first pressure disturbance confirms the existence of gas connectivity, and the reverse pressure disturbance does not meet the above confirmation conditions, the corresponding bubble region pair will not be used to form the final connection edge, and the bubble region pair will be retained as the verification object.

[0169] In embodiments where high detection efficiency is required and calibration has proven that step S1043 can meet the corresponding judgment requirements, the reverse pressure disturbance verification can be skipped, and step S1045 can be executed directly using the gas connectivity relationship obtained in step S1043.

[0170] Step S1045: Determine the connected bubble group based on the gas connectivity relationship; specifically, take each candidate bubble region as a member node, and establish a connection edge between two member nodes that are directly determined in step S1043 or verified in step S1044 to form a bubble connectivity relationship.

[0171] Starting from any member node that has not yet been grouped, read all member nodes directly connected to that member node, and then continue reading member nodes directly connected to the newly joined member node that have not yet been joined to the current group. Repeat the above reading and joining process until there are no more member nodes that can be joined to the current group. Then, determine all the member nodes that have been joined as the same connected bubble group.

[0172] For example, if there is a connecting edge between candidate bubble region A and candidate bubble region B, and a connecting edge between candidate bubble region B and candidate bubble region C, then A, B, and C are grouped into the same connected bubble group, and it is not required that a separate connecting edge be established between A and C.

[0173] Therefore, step S1045 determines visually separated but connected bubble groups through the gas channel below the label based on the gas connectivity relationship. The connected bubble groups are used by step S105 to generate label bubble recognition results according to the group-level objects.

[0174] In step S104, the hidden gas connectivity is determined by the contraction-expansion response during the pressure application phase, the recovery-reverse recovery response during the pressure release phase, the corresponding response time difference, and the response amplitude. Furthermore, multiple visually separated candidate bubble regions that have direct or indirect gas connectivity are grouped into a connected bubble group.

[0175] For example, based on the morphological changes and timestamps of candidate bubble region A and candidate bubble region B obtained in step S103, assuming that candidate bubble region A first forms an effective contraction response at 1.00s and candidate bubble region B first forms an effective expansion response at 1.08s, then the time difference between their pressure response is 1.08s - 1.00s = 0.08s. If the maximum allowable time difference between pressure response is 0.20s, then 0.08s falls within the allowable time range of 0 to 0.20s. After the pressure is released, assuming that candidate bubble region A begins to form an effective recovery response at 2.00s and candidate bubble region B begins to form an effective reverse recovery response at 2.11s, then the recovery response time difference is 2.11s - 2.00s = 0.11s. If the maximum allowable recovery response time difference is 0.25s, then 0.11s falls within the allowable time range of 0 to 0.25s. Furthermore, the contraction response amount of 0.25, expansion response amount of 0.20, recovery response amount of 0.228, and reverse recovery response amount of 0.182 obtained in step S103 are used to calculate the linkage response intensity. Assuming that the effective linkage response amplitude threshold for the pressure stage and the effective linkage response amplitude threshold for the recovery stage are both 0.15, and the response evaluation weight for the pressure stage and the response evaluation weight for the recovery stage are both 0.5, then the corresponding linkage response intensity is 1. If the preset linkage response intensity threshold is 0.70, then the linkage response intensity is not less than 0.70, and candidate bubble region A and candidate bubble region B simultaneously satisfy the response direction condition for the pressure stage, the response direction condition for the recovery stage, and the two response time conditions, thus determining that there is a gas connection relationship between the two. In the implementation of the reverse pressure disturbance verification, candidate bubble region B is then pressured as a new target bubble region. If candidate bubble region A subsequently generates an expansion response and a reverse recovery response after the pressure is released, and the corresponding response time difference and linkage response intensity still meet the gas connectivity judgment requirements, then candidate bubble region A and candidate bubble region B are confirmed to be connected through the gas channel below the label, and a connection edge is established between their corresponding member nodes. If it is further detected that there is also a gas connectivity relationship between candidate bubble region B and candidate bubble region C, then even if a separate connection edge is not established between candidate bubble region A and candidate bubble region C, A, B, and C are merged into the same connected bubble group according to the indirect connection relationship of ABC, thereby obtaining a connected bubble group for group-level evaluation in step S105.

[0176] Step S105: Generate label bubble recognition results based on the overall coverage of the connected bubble group, and determine the adhesion quality status of the label cap based on the label bubble recognition results; the detailed implementation steps of step S105 include: Step S1051: Determine the overall coverage area of ​​the connected bubble group; specifically, for the group containing... Connected bubble groups of candidate bubble regions Read the region boundaries of all candidate bubble regions in the group before pressure is applied, and combine the actual location points corresponding to the boundaries of each region to form a group-level boundary point set.

[0177] In one implementation, convex hull calculation can be performed on the set of group-level boundary points to determine the smallest convex polygon containing all the boundary points of the set of group-level boundary points as the group-level outer envelope region.

[0178] In another implementation, the minimum bounding rectangle containing all member candidate bubble regions can be determined based on the minimum horizontal coordinate, maximum horizontal coordinate, minimum vertical coordinate, and maximum vertical coordinate in the group-level boundary point set, and the minimum bounding rectangle can be used as the group-level outer envelope region.

[0179] The group-level outer envelope region is used to characterize the overall spatial coverage of the connected bubble group on the label surface, and the area of ​​the group-level outer envelope region is not used to replace the actual bubble area.

[0180] The total visible bubble area at the group level is obtained by summing the baseline areas of each member's candidate bubble region before pressure is applied:

[0181] in, Indicates a connected bubble group The total area of ​​visible bubbles at the group level; Indicates the number of candidate bubble regions within a connected bubble group; Indicates the number within the group The baseline area of ​​each candidate bubble region before pressure is calculated; the summation range is all member candidate bubble regions in the current connected bubble group.

[0182] Furthermore, the maximum span at the group level is determined based on the boundary positions of all candidate bubble regions within the connected bubble group:

[0183] in, Indicates a connected bubble group The maximum span of the group level; This represents the set of regions consisting of the actual locations of all candidate bubble regions within the current connected bubble group. and These represent any two actual location points within the set of regions.

[0184] Hidden gas channels in the labeled image that do not meet the criteria for candidate bubble recognition are not directly included. This avoids manually estimating the area of ​​unobservable bubbles when the actual width and height of the hidden gas channels cannot be determined from the images; the group-level spatial relationship formed by the hidden gas channels is characterized by the group-level outer envelope region and the maximum group-level span between member candidate bubble regions.

[0185] Step S1051 determines the overall coverage area based on the connected bubble group. The overall coverage area includes at least the group-level outer envelope region, the total area of ​​visible bubbles at the group level, and the maximum span at the group level, for step S1052 to generate the tag bubble recognition result.

[0186] Step S1052: Generate tag bubble recognition results based on the overall coverage area; specifically, for each connected bubble group, record the connected bubble group identifier, the member candidate bubble regions it contains, the actual position of each member candidate bubble region, the group-level outer envelope region, and the total area of ​​visible bubbles at the group level in the tag bubble recognition results. Maximum span of group level .

[0187] For independent candidate bubble regions that have not established gas connectivity with other candidate bubble regions, the candidate bubble region is treated as an independent bubble identification object, and its actual location, area and maximum span are recorded in the label bubble identification results.

[0188] The maximum span of an independent candidate bubble region is determined by the maximum distance between any two actual locations on the boundary of the corresponding candidate bubble region.

[0189] Therefore, step S1052 generates a tag bubble recognition result based on the overall coverage of the connected bubble group, and the tag bubble recognition result is used by step S1053 to perform a quality threshold comparison.

[0190] Step S1053: Determine the adhesion quality of the label cap based on the label bubble recognition result; specifically, as follows... Figure 3 As shown, for each connected bubble group, the total visible bubble area at the group level is calculated. With bubble area allowable threshold Compare and determine the maximum span of their group levels. With bubble span allowable threshold Compare.

[0191] when: ,or: When this happens, the corresponding connected bubble group is determined to exceed the allowable range for label bubble defects.

[0192] For independent candidate bubble regions that are not assigned to a connected bubble group, their actual region area is compared with the allowable bubble area threshold, and their maximum region span is compared with the allowable bubble span threshold.

[0193] When any connected bubble group or any independent candidate bubble area exceeds the corresponding allowable threshold, the bonding quality of the label cap for label bubble defects is determined to be unqualified.

[0194] When all connected bubble groups and all independent candidate bubble areas do not exceed the corresponding allowable threshold, the bonding quality of the label cap for label bubble defects is determined to be qualified.

[0195] Bubble area allowable threshold And bubble span allowable threshold The information can be read from the existing fit quality requirements for the corresponding product. When the product quality requirements do not directly specify the allowable threshold for connected bubble groups, the total area of ​​visible bubbles at the corresponding group level and the maximum span at the group level can be calculated using confirmed qualified and unqualified samples, respectively. The initial threshold can then be determined based on the statistical distribution of the two types of samples, and finally verified using quality verification samples.

[0196] Through the dimensional calibration in step S1011, and Use the same area unit, and make Using the same length unit as v ensures that the units of comparison are consistent with the units of comparison threshold.

[0197] Therefore, step S1053 determines the bonding quality status of the label cap for label bubble defects based on the label bubble recognition result, and uses the bonding quality status as the final output result of the method of this application.

[0198] In step S105, the gas connectivity structure determined in step S104 is used as the overall evaluation object, instead of continuing to evaluate each visually separated candidate bubble region in the group as a single small bubble. At the same time, the area of ​​the unobservable hidden gas channels in the label image is not artificially estimated, so that the final label bubble recognition result and the bonding quality are based on the detectable candidate bubble region data and the gas connectivity relationship verified by pressure perturbation.

[0199] For example, based on the connected bubble group G1 obtained in step S104, the connected bubble group G1 includes candidate bubble region A and candidate bubble region B, whose reference areas before pressure application are 1.80 mm² and 1.70 mm², respectively. Therefore, the total visible bubble area of ​​the connected bubble group G1 is 1.80 mm² + 1.70 mm² = 3.50 mm². Further, all boundary points of candidate bubble region A and candidate bubble region B are read, and the convex hull method is used to determine the group-level outer envelope region that can surround the two candidate bubble regions. The maximum span of the group is obtained based on the maximum actual distance between all boundary points of the two member candidate bubble regions; assuming the calculated maximum span of the group is 4.80 mm. Step S1052 records "connected bubble group G1, member candidate bubble regions A and B, total visible bubble area of ​​the group 3.50 mm², maximum span of the group 4.80 mm, and corresponding outer envelope region" in the tag bubble recognition result. For the candidate bubble region C identified in step S101, since step S104 did not determine that it has a gas connection with other candidate bubble regions, candidate bubble region C is treated as an independent bubble identification object. Assuming its area is 1.20 mm² and its maximum span is 1.40 mm, it is also recorded in the label bubble identification result. If the allowable threshold for bubble area for the corresponding product is 3.00 mm² and the allowable threshold for bubble span is 4.50 mm, then the total visible bubble area of ​​the connected bubble group G1 (3.50 mm²) is greater than 3.00 mm², and the maximum span (4.80 mm) is greater than 4.50 mm. Therefore, the connected bubble group G1 is determined to exceed the allowable range for label bubble defects, and the current labeling cap's bonding quality for label bubble defects is ultimately determined to be unqualified.

[0200] It should be noted that, before formal testing, this embodiment can establish a set of calibration samples corresponding to the corresponding label material, adhesive layer material, cover type, imaging conditions, and pressure application structure.

[0201] The calibration samples include at least pairs of candidate bubble regions with confirmed gas connectivity and pairs of candidate bubble regions with confirmed gas non-connectivity. The true gas connectivity values ​​corresponding to the calibration samples can be obtained during the calibration phase based on sample preparation records, peeling verification results, or other reference detection results that can confirm whether a gas connectivity relationship exists between bubble regions. These true gas connectivity values ​​are only used to form preset items; manual judgment of whether a gas connectivity relationship exists between candidate bubble regions is not required during the formal detection phase.

[0202] Preset spatial proximity threshold The recoverable pressure range is determined based on the statistical results of the minimum boundary distance of the confirmed gas connectivity calibration samples, and verified using confirmed gas non-connectivity samples. The recoverable pressure range is determined by progressively changing the disturbance pressure value and comparing the effective morphological changes during the pressure application phase and the rate of change of residual area after pressure release at each pressure level.

[0203] Maximum permissible pressure response time difference The maximum permissible pressure response time difference is determined based on the statistical distribution of the pressure response time difference corresponding to the confirmed gas connectivity samples. For example, the 95th percentile of the corresponding pressure response time difference can be used as the initial maximum permissible pressure response time difference. Maximum permissible recovery response time difference. The time difference of recovery response corresponding to the confirmed gas connectivity samples was determined in the same way.

[0204] Effective linkage response amplitude threshold during the pressure phase The threshold is determined based on the statistical results of the non-interlocking area change rate generated during pressurization of confirmed gas-disconnected samples. For example, the mean of the non-interlocking area change rate plus twice the standard deviation can be used as the initial threshold. The effective linkage response amplitude threshold during the recovery phase is also determined. The method was used to determine the non-connected area change rate of the confirmed gas-disconnected sample during the pressure release phase.

[0205] Preset linkage response strength threshold The intensity distribution of the linkage response is determined jointly by the confirmed gas-connected samples and the confirmed gas-disconnected samples. When changes occur in the label material, adhesive layer material, cover curvature, imaging magnification, image acquisition position, or pressure contact end size, causing the existing calibration parameters to no longer meet the corresponding detection requirements, the calibration process corresponding to the changed conditions is re-executed.

[0206] like Figure 4 As shown, this application also provides a label bubble recognition system, including an image acquisition device, a pressure application component, and a processing device.

[0207] The image acquisition device is configured to have an imaging field of view that covers the label area of ​​the label cap to be inspected, for acquiring label images before pressure disturbance, and for acquiring response label images during and after the pressure application component applies local recoverable pressure disturbance to the target bubble area.

[0208] The pressure application component is positioned to apply pressure to the surface of the label to be inspected, and includes a driving component, a pressure contact end, and a pressure detection component. The driving component drives the pressure contact end to move towards or away from the label surface corresponding to the target bubble area; the pressure detection component collects the contact force or directly outputs the pressure detection value. When the pressure contact end adopts a transparent contact structure, the imaging optical path of the image acquisition device can pass through the corresponding transparent area to acquire an image of the target bubble area.

[0209] The processing unit is connected to the image acquisition unit and the pressure application component, respectively.

[0210] The processing device reads the label image acquired by the image acquisition device, identifies multiple separate candidate bubble regions, and determines the spatial proximity relationship between the candidate bubble regions.

[0211] The processing device determines the target bubble region based on the candidate bubble region and its spatial proximity, and outputs pressure disturbance control information to the pressure application component, so that the pressure application component applies a local recoverable pressure disturbance to the target bubble region.

[0212] The processing device further reads the response tag images during and after the pressure disturbance, determines the morphological changes of the target bubble region and other candidate bubble regions, calculates the response time difference and linkage response intensity, determines the gas connectivity relationship based on the corresponding linkage response relationship, and forms a connected bubble group based on the gas connectivity relationship.

[0213] The processing device reads the candidate bubble regions of the connected bubble group and their region boundaries, determines the group-level outer envelope region, the total area of ​​visible bubbles at the group level, and the maximum span at the group level, forms the label bubble recognition result, and determines the bonding quality status of the label cap for label bubble defects based on the label bubble recognition result.

[0214] The processing device may include a processor and a memory. The memory is used to store image data, size calibration relationships, pressure parameters, preset thresholds, and program instructions required to execute the above processing procedures. The processor reads the program instructions and corresponding data to execute the above image analysis, pressure control, response calculation, and quality status judgment processes.

[0215] The formal testing process can be executed automatically after the equipment calibration and parameter initialization are completed. Image size calibration, establishment of recoverable pressure range, establishment of response judgment threshold, and writing of product quality allowable threshold can be completed during equipment initialization or product changeover. During the formal operation phase, image acquisition, candidate bubble identification, target area determination, pressure disturbance, response image analysis, determination of connected bubble groups, and output of bonding quality status can all be executed automatically by the system.

[0216] When the pressure detection value exceeds the upper limit of the recoverable pressure range, the processing device marks the corresponding pressure disturbance as an invalid disturbance and controls the pressure application component to release the pressure. When the residual area change rate of the target bubble region is still greater than the preset recovery threshold after the pressure is released, the current pressure disturbance result is not used to determine gas connectivity, and the pressure recovery anomaly information corresponding to the current candidate bubble region is output for subsequent verification.

[0217] The above description is merely a preferred embodiment of the present invention. It should be noted that, due to the limitations of written expression, there are objectively infinite specific structures. For those skilled in the art, several improvements, modifications, or variations can be made without departing from the principle of the present invention, and the above technical features can be combined in an appropriate manner. These improvements, modifications, variations, or combinations, or the direct application of the concept and technical solution of the present invention to other occasions without modification, should all be considered within the scope of protection of the present invention.

Claims

1. A method for identifying bubble marks on a label cap, characterized in that, include: Acquire a label image of the label cap, identify multiple separate candidate bubble regions, and determine the spatial proximity relationship based on the spatial position between each candidate bubble region; A local recoverable pressure perturbation is applied to the target bubble region in the candidate bubble region, and a response label image is acquired during the pressure perturbation and after the pressure is released; The shape changes of the target bubble region are determined based on the response label image, and other candidate bubble regions that are spatially separated from the target bubble region are filtered based on the spatial proximity relationship. The corresponding shape changes of the filtered other candidate bubble regions are then determined. Based on the linkage response relationship between the target bubble region and the other candidate bubble regions caused by the pressure disturbance, a group of connected bubbles that are visually separated but connected by the gas channel below the label is determined. The label bubble recognition result is generated based on the overall coverage of the connected bubble group, and the adhesion quality status of the label cap is determined based on the label bubble recognition result.

2. The method for identifying bubble marks on a label cap according to claim 1, characterized in that, The label image is segmented into bubble regions to obtain multiple unconnected candidate bubble regions, and the region boundaries of each candidate bubble region are determined. Calculate the minimum boundary distance between any two candidate bubble regions, and establish the spatial proximity relationship between two candidate bubble regions whose minimum boundary distance is not greater than a preset spatial proximity threshold.

3. The method for identifying bubble marks on a label cap according to claim 2, characterized in that, The target bubble region is determined from candidate bubble regions that have at least one spatially adjacent candidate bubble region and whose connectivity determination has not yet been completed; Apply pressure to the target bubble region according to the calibrated recoverable pressure range and maintain the preset disturbance time; The response tag images are acquired according to a preset acquisition sequence before, during and after pressure application, and during the recovery period after pressure is released, forming a disturbance response image sequence corresponding to the target bubble region.

4. The method for identifying bubble marks on a label cap according to claim 3, characterized in that, The positional registration is performed between each response label image in the disturbance response image sequence and the label image before pressure was applied; Extract at least one of the following at each acquisition time: area, equivalent size, and boundary position of the target bubble region and the other candidate bubble regions that have the spatial proximity relationship with the target bubble region and are separated from each other in the label image; The corresponding morphological changes are determined based on the changes in the corresponding parameters at each sampling time relative to before pressure application.

5. The label bubble recognition method according to claim 4, characterized in that, Determine the contraction response of the target bubble region and the expansion response of the remaining candidate bubble regions during the pressure application phase, and determine the recovery response of the target bubble region and the reverse recovery response of the remaining candidate bubble regions after the pressure is released; Determine the time difference between the contraction response and the expansion response, and the time difference between the recovery response and the reverse recovery response; When both the pressure response time difference and the recovery response time difference are within the corresponding allowable time range, the linkage response intensity is determined based on the amplitude of each response, and the gas connectivity relationship is determined based on the linkage response intensity.

6. The label bubble recognition method according to claim 5, characterized in that, After determining that any of the target bubble regions is connected to at least one of the remaining candidate bubble regions through the gas channel below the label, each candidate bubble region is taken as a member node, and the candidate bubble region pairs that are determined to have a gas connection relationship are taken as connecting edges to construct a bubble connection relationship. All member nodes that are directly or indirectly connected through the connecting edges are merged into the same connected bubble group.

7. The method for identifying bubble marks on a label cap according to claim 6, characterized in that, Based on the overall coverage of the connected bubble group, a tag bubble recognition result is generated, including: Obtain the region boundaries of each candidate bubble region within the connected bubble group, and determine the group-level outer envelope region surrounding each member candidate bubble region based on all the region boundaries. The total area of ​​visible bubbles at the group level is determined by summing the areas of the candidate bubble regions of each member, and the maximum span at the group level is determined by the maximum actual distance between the candidate bubble regions of each member; the outer envelope region of the group level, the total area of ​​visible bubbles at the group level, and the maximum span at the group level are recorded in the tag bubble recognition result.

8. The method for identifying bubble marks on a label cap according to claim 7, characterized in that, Determining the adhesion quality of the label cap based on the label bubble recognition result includes: The total visible bubble area and maximum span of each connected bubble group are compared with the bubble area allowable threshold and bubble span allowable threshold, respectively. For candidate bubble regions not included in the connected bubble groups, corresponding comparisons are performed according to their region area and maximum region span. When the total area of ​​visible bubbles in any of the connected bubble groups or the maximum span of the group exceeds the corresponding allowable threshold, or when the area or maximum span of any candidate bubble region not included in the connected bubble group exceeds the corresponding allowable threshold, the bonding quality of the label cap for label bubble defects is determined to be unqualified; otherwise, it is determined to be qualified.

9. A method for identifying bubble marks on a label cap according to claim 5, characterized in that, After determining that the target bubble region has a gas communication relationship with any of the other candidate bubble regions based on the linkage response intensity, the other candidate bubble regions are treated as new target bubble regions and local recoverable pressure disturbances are applied. When the original target bubble region expands during the pressure phase of the new target bubble region and generates a reverse recovery response after the pressure is released, and the corresponding pressure response time difference and recovery response time difference are both within the corresponding allowable time range, and the linkage response intensity determined according to the amplitude of each response reaches the preset linkage response intensity threshold, the two candidate bubble regions are determined to be connected through the gas channel below the label.

10. A label bubble recognition system for label caps, characterized in that, A method for recognizing bubble marks on a label cap as described in any one of claims 1 to 9, comprising an image acquisition device, a pressure application component, and a processing device; The image acquisition device is used to acquire the label image of the labeling cap, and to acquire the response label image during the application of local recoverable pressure disturbance to the target bubble area by the pressure application component and after the pressure is released; The pressure application component is used to apply a localized recoverable pressure perturbation to the target bubble region within the candidate bubble region; The processing device is used to identify multiple mutually separated candidate bubble regions from the label image and determine their spatial proximity based on their spatial location; to determine the morphological changes of the target bubble region and the remaining candidate bubble regions spatially separated from it based on the response label image; to determine the connected bubble group connected through the gas channel below the label based on the linkage response relationship generated by the pressure disturbance; and to generate a label bubble recognition result and determine the adhesion quality status of the label cap based on the overall coverage of the connected bubble group.

Images