A method and system for monitoring the quality state in the production process of an advertisement

Abstract

The application discloses a kind of monitoring method and system of quality state in advertisement production process, it is related to advertisement production intelligent manufacturing technical field, including type determination to boundary according to material flow direction, update three-component state vector by drift rate change and define replica sample, calculate coincidence degree and diversity ratio, update three-component state vector again, obtain characteristic vector by matrix multiplication in combination with fixed constant matrix, calculate boundary quality energy, obtain boundary item, use boundary determination type to revise characteristic vector, obtain closed characteristic vector;By inflow and outflow type marking and state vector structured modeling to process handover boundary, realize the accurate flow tracking and time-series expression of material quality state in multi-process continuous production, improve monitoring stability and anti-interference ability through state flip determination and resurrection resampling mechanism.

Description

Technical Field

[0001] This invention relates to the field of intelligent manufacturing technology for advertising production, and in particular to a method and system for monitoring the quality status during the advertising production process. Background Technology

[0002] In the advertising production process, quality monitoring relies on industrial vision inspection, online sensor acquisition, and PLC process control. It can collect color difference, registration accuracy, and surface defects in real time and make preliminary judgments. Combined with time-series data recording, threshold alarms, and simple statistical analysis, it can judge the pass / fail status of single-frame images or single-point sensor data.

[0003] However, existing technologies still have shortcomings. They monitor quality based on single processes, lacking structured modeling of process handover boundaries, and cannot distinguish between inflow and outflow relationships. Quality information is difficult to form a continuous and traceable time-series quality expression between upstream and downstream processes. Existing methods are based on single frames for judgment, and do not identify and correct state reversals, transient jitter, and positioning jumps. They are prone to misjudging short-term interference as real anomalies, resulting in insufficient stability and anti-interference ability of monitoring results. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a method and system for monitoring the quality status during the advertising production process. It solves the problems of existing technologies that monitor quality based on a single process, lack structured modeling of process handover boundaries, cannot distinguish between inflow and outflow relationships, and make it difficult to form a continuous and traceable time-series quality expression between upstream and downstream processes. Existing methods are based on single frames for judgment and do not identify and correct state reversals, transient jitter, and positioning jumps. They are prone to misjudging short-term interference as real anomalies, resulting in insufficient stability and anti-interference ability of monitoring results.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] In a first aspect, the present invention provides a method for monitoring the quality status during the production process of advertising, comprising the following steps:

[0008] Assign boundary numbers to the process handover points according to the process sequence, collect the coordinates and color difference of the alignment points through an industrial camera, construct and update a three-component state vector, define the replica sample, calculate the overlap and diversity ratio, update the updated three-component state vector again, and perform matrix multiplication with a fixed constant matrix to obtain the feature vector.

[0009] The boundary mass energy is calculated based on the feature vector to obtain the boundary term. The feature vector is corrected using the boundary judgment type to obtain the closed feature vector. Stable candidate samples are constructed based on the boundary term and a stable region is defined. The closed feature vector is compared with the stable region to output the defect label. The source boundary is located by combining the boundary and adjustment instructions are generated.

[0010] As a preferred embodiment of the method for monitoring the quality status during the advertising production process described in this invention, the step of obtaining the feature vector by performing matrix multiplication with a fixed constant matrix includes:

[0011] Based on the lateral drift rate in the three-component state vector, determine whether to flip and obtain the updated three-component state vector;

[0012] Define the frame index as the center of the time subdomain, define the set of frame indices covered by the subdomain as, extract the update three-component vector of each frame in the set of frame indices and define it as a copy sample, calculate the overlap of each copy sample, determine whether to trigger the resurrection event, and obtain the update three-component vector.

[0013] Define a fixed constant matrix, and perform matrix multiplication on the fixed constant matrix and the updated three-component vector to obtain the eigenvector.

[0014] As a preferred embodiment of the method for monitoring the quality status during the advertising production process described in this invention, the step of constructing stable candidate samples and defining a stability region based on boundary terms includes:

[0015] The boundary term is obtained by squaring the three feature components and the defect state scalar in the feature vector of each boundary and summing the squared values.

[0016] The initial start flag is set and the start flag rules are defined to obtain the closed eigenvector. The stable region is then defined by combining the boundary terms.

[0017] As a preferred embodiment of the method for monitoring the quality status during the advertising production process described in this invention, the step of comparing the closed feature vector with the stable region and outputting defect labels includes:

[0018] The closed feature vector is compared with the stable region, and the label inside or outside the region is output. If it is marked as outside the region and the boundary terms of consecutive frames are all greater than 0, it is marked as structural degradation; otherwise, it is marked as temporary anomaly or recoverable perturbation.

[0019] As a preferred embodiment of the method for monitoring the quality status during the advertising production process described in this invention, the step of combining boundary location of the source boundary and generating adjustment instructions includes:

[0020] The boundary numbers of structural degradation are sorted in ascending order, and the boundary with the smallest number is selected as the source boundary. Anomalies are determined for the feature vectors after the source boundary is closed, and adjustment instructions are generated.

[0021] As a preferred embodiment of the method for monitoring the quality status during the advertising production process described in this invention, the construction of the three-component state vector includes:

[0022] The chromatic difference in the boundary observation vector is normalized to obtain the mass intensity component;

[0023] The drift rate is calculated using the coordinates of the positioning points in the boundary observation vector. The mass strength component and the drift rate are arranged vertically to obtain a three-component state vector.

[0024] As a preferred embodiment of the method for monitoring the quality status during the advertising production process described in this invention, the step of assigning boundary numbers to the process handover points according to the process sequence, and collecting the coordinates and color difference of the alignment points using an industrial camera, includes:

[0025] The process sequence is obtained through the API interface, and boundary numbers are assigned to each process handover point from upstream to downstream. For each boundary, the color difference and registration point coordinates are obtained through the API interface.

[0026] Secondly, the present invention provides a quality status monitoring system during the advertising production process, comprising:

[0027] The boundary determination module is used to mark the boundary as inflow or outflow according to the process flow direction and determine the boundary determination type.

[0028] The state update module is used to determine the flip based on the lateral drift rate, correct the state vector, and extract the copy sample.

[0029] The quality calculation module is used to calculate the vector norm and overlap, determine the direction mismatch and execute the revival event to obtain the feature vector.

[0030] The stability region construction module is used to calculate the boundary mass energy and boundary terms, process the start flag, construct the stability region, and maintain the sample queue.

[0031] The anomaly control module is used to determine whether the feature vector is inside or outside the domain, identify structural degradation, locate the source boundary, and output adjustment instructions.

[0032] Thirdly, the present invention provides a computer device including a memory and a processor, wherein the memory stores a computer program, wherein when the computer program is executed by the processor, it implements any step of the method for monitoring the quality status during the advertising production process as described in the first aspect of the present invention.

[0033] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the method for monitoring the quality status during the advertising production process as described in the first aspect of the present invention.

[0034] The beneficial effects of this invention are as follows: This invention determines the type of the boundary according to the material flow direction, updates the three-component state vector and defines replica samples by changing the drift rate, calculates the overlap and diversity ratio, updates the updated three-component state vector again, performs matrix multiplication with a fixed constant matrix to obtain the feature vector, calculates the boundary mass energy to obtain the boundary term, and uses the boundary determination type to correct the feature vector to obtain the closed feature vector; thus, it realizes accurate flow tracking and time-series expression of material quality status in multi-process continuous production, and improves monitoring stability and anti-interference ability. Attached Figure Description

[0035] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0036] Figure 1 This is a flowchart illustrating the operation of the quality status monitoring method during the advertising production process in Example 1.

[0037] Figure 2 This is a schematic diagram of the quality status monitoring system during the advertising production process in Example 1. Detailed Implementation

[0038] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0039] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0040] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0041] Example 1, referring to Figure 1 This is the first embodiment of the present invention, which provides a method for monitoring the quality status during the production process of advertising, including the following steps:

[0042] S1. Assign boundary numbers to the process handover points according to the process sequence, collect the coordinates and color difference of the alignment points through an industrial camera, construct a three-component state vector and update it, define the replica sample, calculate the overlap and diversity ratio, update the updated three-component state vector again, and perform matrix multiplication with a fixed constant matrix to obtain the feature vector.

[0043] Specifically, boundary numbers are assigned to process handover points according to the process sequence, and the coordinates and color difference of the alignment points are collected using an industrial camera, including:

[0044] When the PLC detects a trigger edge, it increments its internal counter by 1 to obtain the frame index k and the timestamp of the frame index k for this acquisition.

[0045] The temperature, humidity, dust, and electrostatic readings of frame index k are obtained through the API interface and arranged horizontally to obtain a random perturbation vector.

[0046] Obtain the process sequence through the API interface, such as printing → drying → lamination → cutting. Assign boundary number b to each process junction from upstream to downstream. For example, the junction of the printing exit and the drying inlet is a process junction.

[0047] After the frame index k trigger edge arrives, for each boundary b, the color difference, the coordinates of the registration point are obtained through the API interface (the image of boundary b is obtained through an industrial camera, and a template matching positioning method (such as normalized cross-correlation NCC) is used. The image and the standard template are used to calculate the normalized cross-correlation score of each coordinate in the image and sort them in descending order. The maximum normalized cross-correlation score is selected and set as the coordinates of the registration point. The standard template is a local image of the material of boundary b that is defect-free and correctly registered) and the defect connected component count are obtained (the image is divided into a binary image using an adaptive threshold segmentation method. All connected regions are extracted from the binary image using connected component analysis. Connected regions with an area greater than or equal to the area threshold are selected and counted) and arranged horizontally to obtain the boundary observation vector.

[0048] The area threshold is used to distinguish between effective defect connected regions and background noise connected regions. It is set based on the imaging resolution of the industrial camera at the corresponding boundary, the minimum perceptible defect size of the material, and the maximum connected region area of ​​normal texture in the template image. The area threshold is limited to being greater than the maximum area of ​​normal texture connected regions and less than the minimum projected area of ​​a single obvious defect in the pixel domain. It is preferably limited to the range of ten to several hundred pixels to ensure that only defect areas that have a real impact on quality are counted.

[0049] By establishing frame indexes and timestamps under PLC triggering synchronization, and combining this with API to obtain environmental parameters, material batches, and process topology in real time, a unique boundary number is assigned to each process handover point. This achieves structured modeling of the quality information flow in the advertising production line. The alignment and positioning points adopt a template matching method based on normalized cross-correlation NCC, using a defect-free standard image as a template to ensure that the positioning accuracy is not affected by surface texture. The defect connectivity counting is effectively eliminated by adaptive threshold segmentation and area filtering, focusing on the real defect area. This design upgrades the traditional isolated single-point detection to a multi-dimensional observation system with boundaries as units, integrating color difference, positioning deviation, and defect density under a unified spatiotemporal coordinate to form a traceable boundary observation vector. Compared with existing technologies that only focus on the internal state of a single process, this solution is the first to explicitly model the process handover interface as a quality transfer node, solving the problem of upstream and downstream quality information breakage and laying a data foundation for subsequent time series analysis and causal inference.

[0050] Furthermore, a three-component state vector is constructed, including:

[0051] The chromatic difference in the boundary observation vector is normalized to obtain the mass intensity component;

[0052] The pixel displacement in the feeding direction is obtained by subtracting the coordinates of the positioning point of frame index k-1 from the coordinates of the positioning point of frame index k in the boundary observation vector. This displacement is then divided by the difference between the timestamps of frame index k and frame index k-1 to obtain the drift rate (including horizontal and vertical coordinates, corresponding to the horizontal and vertical coordinates of the positioning point).

[0053] The mass strength component and drift rate are arranged vertically to obtain a three-component state vector (in column vector form).

[0054] The constructed three-component state vector normalizes the color difference into a quality intensity component, and calculates the lateral and longitudinal drift rates through the difference in coordinates and timestamps of adjacent frame positioning points, forming a compact state description that includes quality characterization and dynamic material movement behavior. This vector not only reflects static quality attributes but also captures the motion characteristics of materials during the conveying process, realizing quality-motion coupled modeling. Drift rate is chosen instead of absolute displacement because the rate better reflects the stability of the equipment and the tension control effect, and is not sensitive to cumulative errors. Compared with existing technologies that only use single-frame color difference or registration deviation for threshold judgment, this solution introduces temporal derivative information, enabling the system to have early perception capabilities for trend shifts, significantly improving the sensitivity to slow degradation processes, while avoiding false alarms caused by single-frame jitter.

[0055] Furthermore, by combining matrix multiplication with a fixed constant matrix, the eigenvectors are obtained, including:

[0056] Based on the process sequence, if the material enters the boundary b from the upstream process, the boundary type is marked as inflow; if the material flows out from the boundary b to the downstream process, the boundary type is marked as outflow, thus obtaining the boundary determination type.

[0057] Based on the lateral drift rate in the three-component state vector, if the lateral drift rate of frame index k is greater than or equal to 0 and the lateral drift rate of frame index k-1 is less than 0, or the lateral drift rate of frame index k is less than 0 and the lateral drift rate of frame index k-1 is greater than or equal to 0, it is determined to be a flip; otherwise, it is not a flip.

[0058] If the frame is flipped, the lateral drift rate of frame index k is set to the lateral drift rate of frame index k-1. If the frame is not flipped, the lateral drift rate of frame index k is retained, and the updated three-component state vector is obtained.

[0059] Define the frame index k as the center of the time subdomain, define the set of frame indices covered by the subdomain as (k-1, k, k+1, k+2...), extract the update three-component vector of each frame in the set of frame indices and define it as a replica sample;

[0060] Calculate the Euclidean norm of each replica sample, construct a normalized vector, and calculate the overlap ratio using the following formula:

[0061] ,

[0062] ,

[0063] in, For normalized vectors, For dimension (3), It is the Euclidean norm. For the degree of overlap, For transpose, Number the replica sample. For frame indexing;

[0064] If the overlap is less than 0, it is determined to be a direction mismatch (transient reflection / mismatched positioning / severe jitter causing state direction reversal), and a resurrection event is triggered; otherwise, the diversity ratio is calculated.

[0065] The calculation of the diversity ratio includes dividing the number of active components by the dimension (3) to obtain the diversity ratio. If the diversity ratio is less than or equal to one-half, it is determined to be a direction mismatch and a resurrection event is triggered. Otherwise, the updated three-component state vector is retained and subsequent steps are performed.

[0066] The number of active components refers to the threshold determination of each component vector in the three component vectors to be updated, including setting it to 1 if the absolute value of the component vector is greater than or equal to the mass strength threshold / lateral drift rate threshold / longitudinal drift rate threshold, otherwise it is 0. The set values ​​of the three component vectors are summed to obtain the number of active components.

[0067] The quality intensity threshold is used to determine whether the color difference change reaches the level of perceptible quality disturbance. It is set based on the color difference fluctuation range of the standard template image under stable production conditions and the minimum quality deviation that is acceptable to humans or equipment. The quality intensity threshold is limited to being higher than the upper limit of the natural color difference fluctuation in normal production and lower than the color difference amplitude corresponding to obvious defects. It is preferably a low proportion range in the normalized color difference range to avoid misjudging background noise as quality abnormality.

[0068] The lateral drift rate threshold and longitudinal drift rate threshold are used to determine the effective displacement change of the alignment positioning point. The setting basis is the mechanical stability of the conveying mechanism, the minimum resolution of the encoder, and the maximum allowable offset speed of the positioning point under normal material feeding conditions. The drift rate threshold is limited to being higher than the false displacement rate caused by camera resolution and timestamp jitter, and lower than the typical drift rate when the actual process is abnormal. It is preferably limited to a range of several pixels per unit time.

[0069] The resurrection event involves adding another image acquisition to boundary b and re-extracting the coordinates of the registration and positioning point. The above operation process is repeated to obtain the updated three-component vector.

[0070] Define a fixed constant matrix Matrix multiplication is performed on a fixed constant matrix and an updated three-component vector to obtain an eigenvector, which includes three eigencomponents. The formula is as follows:

[0071] .

[0072] By detecting flip events through changes in the sign of the lateral drift rate and reverting to the previous frame state when a mismatch in direction is determined, false state abrupt changes caused by reflections, occlusions, or positioning jumps are effectively suppressed. Furthermore, dual criteria of duplicate samples, overlap, and diversity ratio are introduced to verify state reliability from two dimensions: vector direction consistency and component activity. This breaks through the limitations of traditional moving averages or low-pass filtering that rely solely on numerical smoothing. After the resurrection event is triggered, the image is re-acquired and the state is reconstructed, forming a closed-loop correction mechanism. The design of a fixed constant matrix maps the physically meaningful original state into a more discriminative feature vector, enhancing the sensitivity of features to specific abnormal patterns. The overall process significantly improves the robustness and anti-interference capability of the system in complex industrial environments.

[0073] S2. Calculate the boundary mass energy based on the feature vector to obtain the boundary term. Correct the feature vector using the boundary judgment type to obtain the closed feature vector. Construct stable candidate samples based on the boundary term and define the stable region. Compare the closed feature vector with the stable region to output the defect label. Combine the boundary to locate the source boundary and generate adjustment instructions.

[0074] Specifically, stable candidate samples are constructed and a stable region is defined based on boundary terms, including:

[0075] The boundary mass energy is obtained by taking the squares of the three characteristic components in the characteristic vector of each boundary b and the defect state scalar, and then summing the squares.

[0076] The defect state scalar is the normalized value of the defect connected component count;

[0077] The boundary term is obtained by subtracting the boundary quality energy of frame index k-1 from the boundary quality energy of frame index k.

[0078] The initial start flag is 0 or 1 (marked as 1 when the reference feature vector is generated for the first time, otherwise 0), and the start flag rules are defined, including: if it is 0, the feature vector is defined as the closed feature vector; if it is 1 and it is an inflow, the drift rate in the feature vector is replaced with the drift rate in the reference feature vector (which is the median of each dimension in the queue in subsequent steps); if it is 1 and it is an outflow, the lateral drift rate in the feature vector is replaced with the lateral drift rate in the reference feature vector, thus obtaining the closed feature vector.

[0079] If the boundary term is less than or equal to 0, then frame index k is marked as a stable candidate sample, and the perturbation category number is set as the grouping key. The stable candidate sample is added to the tail of the queue of the grouping key. If the queue length is greater than the queue length threshold, then the stable candidate sample at the head of the queue is deleted, making the queue length equal to the queue length threshold. The samples in the queue are sorted in ascending order, and the sorted lower quartile and upper quartile samples are extracted. The stability region is defined by the formula:

[0080]

[0081] in, For the stability region, Number the boundaries. As a grouping key, The closed feature vector to be determined. , as well as These are the three components of the closed eigenvector to be determined. , as well as The lower bound vector is obtained by vertically concatenating the three components of the closed eigenvector of the lower quartile samples. , as well as The upper bound vector is obtained by vertically concatenating the three components of the closed feature vector of the upper quartile sample.

[0082] The queue length threshold is used to construct the statistical basis for stable candidate samples. It is set based on the stable production cycle, the duration of random disturbances, and the minimum observation window length of boundary state changes. The queue length threshold is limited to the number of frames that can cover a complete stable operating cycle without crossing obvious process switching periods. It is preferably limited to the range of tens to hundreds of frames, so as to avoid the lag effect of historical states on the current judgment while ensuring statistical reliability.

[0083] The disturbance category number refers to determining whether the data in the random disturbance vector that are not material batch numbers are within the allowable range. If they are, the value is 1; otherwise, it is 0. The judgment results are arranged vertically to obtain the physical disturbance pattern vector, and a disturbance category number (physical disturbance pattern vector, material batch number) is constructed.

[0084] The overall degradation degree of the boundary state is quantified by boundary mass energy, and the boundary term is used to determine whether the quality trend has improved or deteriorated. The activation flag rule fuses the reference feature vector differently according to the boundary type, ensuring that the inflow boundary inherits the upstream steady-state information and the outflow boundary retains its own lateral dynamic characteristics, reflecting a deep modeling of the material flow direction. Stable candidate samples are grouped and enqueued according to the disturbance category number, so that the steady-state distribution under different environmental disturbance modes is modeled independently, avoiding the distortion of the stable domain caused by the mixing of heterogeneous working conditions. The stable domain constructed based on quartiles has strong anti-outlier capability and can adaptively reflect the normal fluctuation range under the current disturbance conditions. This mechanism upgrades the static threshold to a dynamic, grouped, and time-series-aware stable interval, which greatly reduces the misjudgment rate caused by environmental fluctuations.

[0085] Furthermore, the closed eigenvectors are compared with the stable region to output defect labels, including:

[0086] The closed feature vector is compared with the stable region, and the label of inside or outside the region is output. If the first dimension of the closed feature vector is between the upper and lower bounds of the first dimension of the stable region, and the second dimension is between the upper and lower bounds, and the third dimension is between the upper and lower bounds, then it is determined to be inside the region; otherwise, it is determined to be outside the region.

[0087] If the boundary terms are marked as being outside the domain and are all greater than 0 for g consecutive frames, then it is marked as structural degradation; otherwise, it is marked as temporary anomaly or recoverable disturbance, and a defect label is obtained.

[0088] By performing a dimension-wise inclusiveness test on the eigenvectors after closure and the stable region, a high-confidence anomaly determination is achieved. Only when all dimensions exceed the boundary is it marked as out of the domain, avoiding oversensitivity caused by fluctuations in a single indicator. Combined with the trend criterion of continuous positive growth of boundary terms, it distinguishes between structural degradation (continuous deterioration) and temporary anomalies (recoverable disturbances), solving the fundamental defect of existing technologies that cannot distinguish between faults and interference. This labeling system not only outputs the existence of anomalies but also provides anomaly type classification, providing semantic basis for subsequent decision-making. Compared with the traditional binary pass / fail determination, this solution achieves refined identification of anomaly types, supporting more intelligent operation and maintenance strategies.

[0089] Furthermore, by combining the boundary location with the source boundary, adjustment instructions are generated, including:

[0090] The boundary numbers of structural degradation are sorted in ascending order, and the boundary with the smallest number is set as the source boundary. If the source boundary is the upper bound of the third dimension of the superstability region of the eigenvector after closure, it is determined that the lateral drift is too large in the positive direction. If it is the lower bound of the third dimension of the superstability region, it is determined that the lateral drift is too large in the negative direction. Otherwise, the third dimension is not selected for judgment, and the first or second dimension is used for judgment. If the first or second dimension exceeds the boundary, it is determined that the strength / material flow coupling is abnormal.

[0091] If the lateral drift is too large in the positive direction, the phase register is subtracted by the fixed phase step (encoder pulse count or minimum angle resolution). If the lateral drift is too large in the negative direction, the phase register is added to the fixed phase step. If the strength / material feeding coupling is abnormal and the source boundary is located at the drying-related interface, the temperature register is increased or decreased according to the temperature step. If the strength / material feeding coupling is abnormal and the source boundary is located at the tension-related interface, the tension register is increased or decreased according to the tension step, thus obtaining the adjustment command.

[0092] By using the structural degradation boundary with the smallest sequence number in ascending order as the source boundary, this approach aligns with the physical laws governing the propagation of defects along the material flow in advertising production lines. This ensures that control commands act on the starting point of the problem. Based on the super-boundary dimension, the abnormal mechanism is automatically identified. The third dimension corresponds to lateral drift and is directly related to phase control. The first and second dimensions reflect the coupling between quality strength and longitudinal material movement, pointing to temperature or tension adjustment. This diagnostic logic based on the semantics of feature dimensions transforms the abstract vector super-boundary into specific equipment parameter adjustment commands, achieving a closed loop from monitoring to control. The phase, temperature, and tension registers are increased or decreased in fixed steps, ensuring the determinism of the adjustment action while avoiding the risk of over-adjustment. This mechanism is the first in the advertising production field to achieve precise process self-tuning based on boundary tracing, significantly outperforming the extensive processing methods of global shutdown or manual intervention in existing technologies.

[0093] Example 2, refer to Figure 2As a second embodiment of the present invention, a quality status monitoring system during the advertising production process includes:

[0094] The boundary determination module is used to mark the boundary as inflow or outflow according to the process flow direction and determine the boundary determination type.

[0095] The state update module is used to determine the flip based on the lateral drift rate, correct the state vector, and extract the copy sample.

[0096] The quality calculation module is used to calculate the vector norm and overlap, determine the direction mismatch and execute the revival event to obtain the feature vector.

[0097] The stability region construction module is used to calculate the boundary mass energy and boundary terms, process the start flag, construct the stability region, and maintain the sample queue.

[0098] The anomaly control module is used to determine whether the feature vector is inside or outside the domain, identify structural degradation, locate the source boundary, and output adjustment instructions.

[0099] This embodiment also provides a computer device applicable to a method for monitoring the quality status during the advertising production process, comprising: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the method for monitoring the quality status during the advertising production process as proposed in the above embodiment.

[0100] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0101] This embodiment also provides a storage medium storing a computer program, which, when executed by a processor, implements the method for monitoring the quality status during the advertising production process as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0102] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for monitoring the quality status during the production process of advertising, characterized in that: Includes the following steps: Assign boundary numbers to the process handover points according to the process sequence, collect the coordinates and color difference of the alignment points through an industrial camera, construct and update a three-component state vector, define the replica sample, calculate the overlap and diversity ratio, update the updated three-component state vector again, and perform matrix multiplication with a fixed constant matrix to obtain the feature vector. The boundary mass energy is calculated based on the feature vector to obtain the boundary term. The feature vector is corrected using the boundary judgment type to obtain the closed feature vector. Stable candidate samples are constructed based on the boundary term and a stable region is defined. The closed feature vector is compared with the stable region to output the defect label. The source boundary is located by combining the boundary and adjustment instructions are generated.

2. The method for monitoring the quality status during the advertising production process as described in claim 1, characterized in that: The process of obtaining the feature vector by combining a fixed constant matrix with matrix multiplication includes: Based on the lateral drift rate in the three-component state vector, determine whether to flip and obtain the updated three-component state vector; Define the frame index as the center of the time subdomain, define the set of frame indices covered by the subdomain as, extract the update three-component vector of each frame in the set of frame indices and define it as a copy sample, calculate the overlap of each copy sample, determine whether to trigger the resurrection event, and obtain the update three-component vector. Define a fixed constant matrix, and perform matrix multiplication on the fixed constant matrix and the updated three-component vector to obtain the eigenvector.

3. The method for monitoring the quality status during the advertising production process as described in claim 2, characterized in that: The construction of stable candidate samples and definition of stable regions based on boundary terms includes: The boundary term is obtained by squaring the three feature components and the defect state scalar in the feature vector of each boundary and summing the squared values. The initial start flag is set and the start flag rules are defined to obtain the closed eigenvector. The stable region is then defined by combining the boundary terms.

4. The method for monitoring the quality status during the advertising production process as described in claim 3, characterized in that: The comparison between the closed feature vector and the stable region, and the output of the defect label, includes: The closed feature vector is compared with the stable region, and the label inside or outside the region is output. If it is marked as outside the region and the boundary terms of consecutive frames are all greater than 0, it is marked as structural degradation; otherwise, it is marked as temporary anomaly or recoverable perturbation.

5. The method for monitoring the quality status during the advertising production process as described in claim 4, characterized in that: The process of combining boundary location to determine the source boundary and generating adjustment instructions includes: The boundary numbers of structural degradation are sorted in ascending order, and the boundary with the smallest number is selected as the source boundary. Anomalies are determined for the feature vectors after the source boundary is closed, and adjustment instructions are generated.

6. The method for monitoring the quality status during the advertising production process as described in claim 5, characterized in that: The construction of the three-component state vector includes: The chromatic difference in the boundary observation vector is normalized to obtain the mass intensity component; The drift rate is calculated using the coordinates of the positioning points in the boundary observation vector. The mass strength component and the drift rate are arranged vertically to obtain a three-component state vector.

7. The method for monitoring the quality status during the advertising production process as described in claim 6, characterized in that: The process of assigning boundary numbers to process handover points according to the process sequence, and collecting the coordinates and color difference of the alignment points using an industrial camera, includes: The process sequence is obtained through the API interface, and boundary numbers are assigned to each process handover point from upstream to downstream. For each boundary, the color difference and registration point coordinates are obtained through the API interface.

8. A quality status monitoring system for the advertising production process, used to implement the quality status monitoring method for the advertising production process as described in any one of claims 1 to 7, characterized in that: include: The boundary determination module is used to mark the boundary as inflow or outflow according to the process flow direction and determine the boundary determination type. The state update module is used to determine the flip based on the lateral drift rate, correct the state vector, and extract the copy sample. The quality calculation module is used to calculate the vector norm and overlap, determine the direction mismatch and execute the revival event to obtain the feature vector. The stability region construction module is used to calculate the boundary mass energy and boundary terms, process the start flag, construct the stability region, and maintain the sample queue. The anomaly control module is used to determine whether the feature vector is inside or outside the domain, identify structural degradation, locate the source boundary, and output adjustment instructions.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the method for monitoring the quality status during the advertising production process as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the method for monitoring the quality status during the advertising production process as described in any one of claims 1 to 7.

Images