Method, apparatus, and system for providing automated marking and inspection solution for surface printing process based on product material

Abstract

One embodiment of the present invention relates to a method, apparatus, and system for providing an automated marking and inspection solution for a surface printing process according to product material, which actively determines an optimal marking method by analyzing the multidimensional surface physical and chemical characteristics of a target product, and performs predictive maintenance by analyzing the topological defect distribution of a marked identification code and logistics environment stress using machine learning to predict the possibility of future degradation.

Description

Technology Field

[0001] The following embodiments relate to a method, apparatus, and system for providing an automated marking and inspection solution for a surface printing process according to product material, which actively determines an optimal marking method by analyzing the multidimensional surface physical and chemical properties of a target product, and performs predictive maintenance by analyzing the topological defect distribution of the marked identification code and logistics environment stress using machine learning to predict the possibility of future degradation. Background Technology

[0002] As smart factories and unmanned logistics systems advance, the reliability of printing and reading identification codes, such as DataMatrix, QR codes, and barcodes, for identifying and tracking products is emerging as a key factor determining the productivity of the entire process.

[0003] Conventional surface marking processes have uniformly applied a single marking method (e.g., batch inkjet ejection or batch laser engraving) without considering various physical and chemical material characteristics of the target product, such as surface roughness, light reflectance, and surface energy (dyne). As a result, fundamental process defects frequently occurred, such as ink failure or capillary bleeding in materials with low surface energy, and omission of engraving or equipment damage caused by laser diffuse reflection in metal materials with high reflectance.

[0004] Moreover, existing vision inspection systems were limited to determining only one-dimensional read rate pass / fail status based on the current contrast of the captured image or simple pixel matching. 2D identification codes had limitations in that the criticality of defects varied completely depending on topological importance, such as finder patterns or quiet zones, and they appeared as merely a simple score drop until the tipping point of the Reed-Solomon error correction algorithm was reached.

[0005] Above all, the most critical limitation of conventional technology is that it excludes the 'logistics environment after shipment from the factory.' Even if a product barely passes vision camera inspection immediately after shipment, exposure to environmental stresses such as friction, vibration, and high temperature and humidity conditions during long-term sea or land transportation causes minute ink smudging or microcracks at the laser etching interface to deteriorate rapidly, eventually transitioning to a 'No-Read' state at the end customer or logistics hub. This results in massive logistics claims and costs for full rework; however, existing fragmentary quality inspection systems lack an integrated feedback control mechanism capable of predicting such future risks of damage over time and avoiding them through safety modes, such as label adhesion. Prior art literature

[0006] (Patent Document 0001) KR 10-2532703 B (Patent Document 0002) KR 10-2257219 B (Patent Document 0003) KR 10-2717167 B The problem to be solved

[0007] The problem that an embodiment of the present invention aims to solve is to provide a method, apparatus, and system for an automated marking and inspection solution for a surface printing process according to product material, which overcomes the limitations of conventional uniform marking application and fragmentary vision reading inspection as described above by dynamically switching heterogeneous marking methods according to the multidimensional surface characteristics of the product, and at the same time predicting the risk of unreadable conversion after shipment (remaining valid time) through a time-series machine learning model that fuses the derived multidimensional defect distribution pattern and logistics transport metadata, and feeding this back to real-time equipment control. means of solving the problem

[0008] According to one embodiment, a method for providing an automated marking and inspection solution for a surface printing process according to a product material, wherein the device includes a processor, memory, a communication module, and a non-transient storage medium, and the processor executes a program stored in the non-transient storage medium; the method comprises: a marking method determination step of acquiring surface material data of a target product entering a production line and determining one of ink marking, laser marking, and label attachment optimized for the surface material data by referring to a pre-established marking mapping table for each material; and a marking execution step of printing or attaching an identification code on the surface of the target product by activating one of a selected mode among a first marking mode that controls the ejection of an inkjet head according to the marking method, a second marking mode that etches the surface by controlling the output and irradiation path of a laser oscillator, and a third marking mode that attaches a physical label by driving a label dispenser. An inspection execution step of acquiring an inspection image by capturing a region of interest (ROI) marked with the identification code using a vision camera, and calculating quality reading data including at least one of contrast, shape distortion, and read rate of the identification code within the inspection image based on a vision reading algorithm; and an integrated management step of mutually mapping the control parameters applied in the marking execution step with the quality reading data and storing them in a history database, and performing automatic feedback control to change the determination logic of the marking method or the control parameters corresponding to the surface material data of the target product when the quality reading data is below a preset allowable threshold.The present invention provides a method for providing an automated marking and inspection solution for a surface printing process according to product material, including: a method for providing an automated marking and inspection solution for a surface printing process according to product material; a method for determining an adaptive marking mode and automating inspection based on a multidimensional characteristic vector of a surface material; a method for inspecting optical readability quality through topological segmentation of an identification code and inference of error correction exhaustion rate; a method for predicting the degradation of an identification code readability rate using a time-series environmental stress transfer matrix; a predictive maintenance marking quality management method based on logistics metadata and calculation of remaining valid time (RUL); and an automatic feedback method for printing process control parameters through multidimensional defect distribution pattern analysis.

[0009] Additionally, the marking method determination step comprises: calculating a surface roughness index, a light reflectance index, a glossiness index, and a surface energy estimate of a target product based on sensing information measured from a sensor unit placed on the production line, and generating a multidimensional characteristic vector including the surface roughness index, the light reflectance index, the glossiness index, and the surface energy estimate; obtaining a marking mode-specific weight matrix, an allowable threshold set, and an exclusion rule set corresponding to each of the first to third marking modes from the material-specific marking mapping table by referring to the material code, item code, required reading grade, and production speed condition included in the product identification information of the target product; and performing matrix operations on the multidimensional characteristic vector and the marking mode-specific weight matrix to calculate a mode-specific basic suitability value including ink wettability suitability for the first marking mode of the target product, laser response suitability for the second marking mode, and label adhesion suitability for the third marking mode. A step of calculating a final process suitability score for each of the first to third marking modes by reflecting, respectively, risk penalty factors including risk of ink bleeding, risk of non-adhesion, risk of laser diffuse reflection, risk of thermal damage, risk of label peeling, and risk of attachment position error for the basic suitability values ​​for each of the above modes; a step of excluding the first marking mode from the determination candidates to preemptively prevent ink drop, bleeding, or non-adhesion defects if the surface energy estimate is less than a preset threshold dyne value for securing ink wettability, or if the surface roughness index deviates from a preset roughness range that hinders uniform ink application.The method may include: a step of excluding the second marking mode from the determination candidates to prevent engraving omission, contrast degradation, or damage to surrounding optical equipment due to diffuse reflection of the laser beam when the light reflectance index or glossiness index is above a preset high-reflection threshold or the thermal sensitivity coefficient corresponding to the material code is above a preset thermal damage criterion; and a mode selection step of selecting the marking mode with the highest final process suitability score as the priority candidate method, wherein if the score difference between the priority candidate method and the next-ranked method is less than a preset competition threshold, additionally calculating a correction score reflecting past reading success rates, defect rates, and rework rates for the same material group or similar surface characteristic groups by referring to a history database, and determining the marking mode with the higher correction score as the final marking method, wherein if both the first marking mode and the second marking mode are excluded from the determination candidates, determining the third marking mode as the final marking method.

[0010] And, the inspection execution step comprises: a step of controlling the irradiation angle, irradiation direction, visible light wavelength band, polarization state, and light intensity of a multi-wavelength illumination module based on an activated marking mode, expected marking location information of a target product, and optical response characteristics of the surface of a target product, and acquiring a diffuse reflection inspection image in which specular reflection components are suppressed; a step of performing lens distortion correction, perspective correction, luminance flattening, highlight removal, and image alignment on the diffuse reflection inspection image, and generating an aligned ROI image by extracting a region of interest marked with an identification code by referencing at least one of the outer contour, reference mark, reference hole, corner, and product reference axis of the target product; and a step of calling an image processing recipe pre-mapped to the activated marking mode and selectively applying at least one morphological filter among an adaptive binarization algorithm, a contrast enhancement filter, an edge enhancement filter, and dilation, erosion, opening, and closing to the aligned ROI image, thereby segmenting the cell area, code boundary area, and background area of ​​the identification code. When the activated marking mode is a first marking mode, the step of measuring the distribution density of satellite droplets, the capillary spreading area between cells, the ratio of unejected blank pixels, the deviation in ink diffusion directionality, and the boundary roughness index of the code outer edge based on the cell area and the code boundary area, and deriving a first mode index including an ink spreading defect index and a shape distortion index using the distribution density of satellite droplets, the capillary spreading area between cells, the ratio of unejected blank pixels, the deviation in ink diffusion directionality, and the boundary roughness index of the code outer edge;When the above-mentioned activated marking mode is a second marking mode, the method comprises the step of measuring the discoloration spectrum deviation of the heat-affected zone, the frequency of occurrence of microcrack candidate patterns at the etching interface, the optical response non-uniformity at the laser beam overlap point, and the contrast deviation corresponding to the carbonized edge width and etching depth using the optical response difference due to multi-angle illumination and the change in reflected luminance by wavelength band, and deriving a second mode index including a laser etching integrity index and a contrast index using the discoloration spectrum deviation of the heat-affected zone, the frequency of occurrence of microcrack candidate patterns at the etching interface, the optical response non-uniformity at the laser beam overlap point, and the contrast deviation corresponding to the carbonized edge width and etching depth; If the activated marking mode is a third marking mode, the method comprises the steps of: performing contour tracking and edge profile analysis along with code outline information required for reading the identification code to measure an end lifting index based on changes in the radius of curvature of the label edge, a label adhesion misalignment angle relative to the product reference axis, an area of ​​bubble occurrence within the adhesive surface, a wrinkle distribution density, and a positional deviation of the label outline; deriving a third mode index including a physical adhesion integrity index using the end lifting index, the label adhesion misalignment angle relative to the product reference axis, an area of ​​bubble occurrence within the adhesive surface, a wrinkle distribution density, and a positional deviation of the label outline; calculating a basic read rate value of the cell area of ​​the identification code based on international standard barcode specifications; generating a defect feature vector by normalizing the detailed defect measurement value derived corresponding to the activated marking mode and the mode index corresponding to the activated marking mode; and calculating a weighted penalty based on the severity of each item included in the defect feature vector. A step of calculating an effective reading rate by reflecting a point deduction weighting penalty to the above-mentioned basic reading rate value;The method may include: a step of analyzing a multidimensional defect distribution pattern from the above defect feature vector to calculate a degradation prediction score that quantifies physical vulnerability to friction, vibration, and temperature and humidity change environments in the logistics post-process; a step of calculating an initial read failure risk based on the difference between the above valid read rate and a preset target read threshold value, and calculating a readability conversion risk by reflecting the degradation prediction score in the initial read failure risk; and a step of generating final quality read data by mapping the above valid read rate, defect feature vector, degradation prediction score, and readability conversion risk.

[0011] In addition, the step of calculating the above-mentioned weighted penalty comprises: aligning a symbolage layout corresponding to the specification information of the identification code onto a reference coordinate system, and then spatially dividing the symbolage layout to establish a plurality of topological importance zones including a finder pattern area, a timing pattern area, a data encoding area, and a quiet zone area, wherein if at least one of an alignment pattern area, a version information area, and a format information area exists according to the identification code specification information, this is set as an additional auxiliary importance zone; and, based on the spatial data of the aligned ROI image and the divided cell area and code boundary area, mapping the pixel coordinates of detailed defect measurements, defect area, defect outline shape, and defect clustering diagram to the reference coordinate system to generate a zone-specific defect distribution matrix including the defect occupancy area, defect occupancy ratio, defect clustering diagram, defect center coordinate distribution, and continuous intrusion length of defects for each of the topological importance zones. A step of identifying an error correction level corresponding to the international standard barcode specification of the identification code, and calculating an error correction exhaustion rate representing the ratio of codewords damaged due to defects relative to the total error correction restoration capacity of the identification code, based on the defect occupancy ratio and the continuous intrusion length of defects within the data encoding area; a step of, for the zone-specific defect distribution matrix, assigning a first weight of the highest grade to defects occurring in the finder pattern area and timing pattern area, reflecting the risk of direct connection to unrecognizability due to loss of decoding reference axis or code alignment failure; assigning a second weight to defects occurring in the quiet zone area, reflecting the level of background separation margin infringement; and assigning an auxiliary weight to defects occurring in the alignment pattern area, version information area, or format information area, reflecting the level of code consistency degradation.For defects occurring in the data encoding area, a step of calculating a third weight by inputting the error correction exhaustion rate, defect clustering, and continuous intrusion length of the defect into a weighting function defined such that the weight increases non-linearly based on a threshold as the error correction exhaustion rate approaches a preset limit restoration rate; a step of calculating a zone-specific penalty component value by combining the corresponding weight, defect occupancy area, and defect occupancy ratio for each of the finder pattern area, timing pattern area, data encoding area, quiet zone area, and auxiliary importance area, and calculating a first penalty value by weighted summing the zone-specific penalty component values; The method may include the step of calculating a synergy coefficient that reflects the reading interference effect due to mutual interference between multiple defects when, within the defect feature vector, at least two detailed defect measurements among the ratio of undischarged blank pixels, the area of ​​inter-cell capillary bleeding, the contrast deviation corresponding to the etching depth, the frequency of occurrence of microcrack candidate patterns at the etching interface, the end lifting index, and the label adhesion misalignment angle relative to the product reference axis overlap across adjacent topological importance zones, or when the clustering of detailed defect measurements that directly impair the contrast of the code exceeds a preset threshold; and the step of deriving a final deduction weighted penalty by reflecting the synergy coefficient in the first penalty value, wherein an additional upper limit coefficient is applied so that the penalty value before and after the application of the synergy coefficient does not exceed a preset stability limit.

[0012] Additionally, the step of calculating the degradation prediction score comprises: extracting a multidimensional defect distribution pattern representing the damage location, damage size, damage density, and damage propagation directionality of the identification code based on the defect feature vector, the defect cluster diagram included in the zone-specific defect distribution matrix, the defect occupancy ratio, and the continuous intrusion length of the defect, the valid read rate, and the deduction weighted penalty; projecting the multidimensional defect distribution pattern onto a pre-established environmental stress transfer matrix to calculate the basic sensitivity for each environmental factor for the friction factor, vibration factor, temperature change factor, humidity change factor, repetitive impact factor, and repetitive contact factor applied to the identification code; and, when the activated marking mode is the first marking mode, up-correcting the basic sensitivity corresponding to the temperature change factor and humidity change factor in proportion to the inter-cell capillary spreading area, the ratio of undischarged blank pixels, and the shape distortion index. When the activated marking mode is a second marking mode, a step of up-correcting the basic sensitivity corresponding to the vibration factor and the repetitive impact factor in proportion to the frequency of occurrence of microcrack candidate patterns on the etching interface, the contrast deviation corresponding to the etching depth, and the laser etching integrity index; when the activated marking mode is a third marking mode, a step of up-correcting the basic sensitivity corresponding to the friction factor and the repetitive contact factor in proportion to the end lifting index, the label adhesion misalignment angle relative to the product reference axis, the bubble generation area within the adhesive surface, and the physical adhesion integrity index; and a step of obtaining environmental stress metadata from a logistics database linked with the identification information of the target product, including the estimated logistics transportation time after shipment of the target product, the dwell time per transportation segment, the estimated accumulated vibration per transportation method, the number of loading and unloading operations, the estimated temperature exposure profile and the estimated humidity exposure profile along the route.A step of generating a time-series stress sequence including friction stress, vibration stress, temperature change stress, humidity change stress, repetitive impact stress, and repetitive contact stress for each of multiple time intervals of the logistics post-processing by dividing the above environmental stress metadata according to chronological order; a step of inferring a degradation prediction curve including a trend of reading rate decline, a trend of contrast reduction, a trend of shape distortion increase, and a point of transition to unreadability according to the elapsed time of the logistics post-processing by inputting the basic sensitivity for each environmental factor and the time-series stress sequence as multiple variables into a machine learning-based time-series degradation prediction model pre-trained using historical data of identification codes judged unreadable in the past logistics environment, a history of changes in the past valid reading rate, a history of the past defect feature vector, and past environmental stress metadata; and a step of calculating, from the degradation prediction curve, the remaining valid time until the reading rate of the identification code drops below a preset limit reading value and the remaining quality maintenance time until the contrast or shape distortion exceeds a preset quality limit. A step of calculating a readability margin value based on the difference between the estimated logistics transport time and the remaining valid time, and calculating a quality maintenance margin value based on the difference between the estimated logistics transport time and the remaining quality maintenance time, and then calculating a readability margin ratio and a quality maintenance margin ratio corresponding to the readability margin value and the quality maintenance margin value, respectively; a step of setting the smaller value among the readability margin ratio and the quality maintenance margin ratio as the representative margin ratio, and inputting the representative margin ratio into a surcharge function defined such that the weight increases non-linearly as the representative margin ratio decreases below a preset safety guarantee threshold to calculate a non-linear surcharge coefficient;and may include the step of finally calculating a degradation prediction score that quantitatively indicates the possibility that the identification code will transition to an unreadable state after the logistics post-processing by combining the slope of the reading rate decline derived from the degradation prediction curve, the remaining effective time, the representative margin ratio, the basic sensitivity by environmental factor, and the non-linear markup coefficient.

[0013] A device according to one embodiment may be combined with hardware and controlled by a computer program stored on a medium to execute the method of any one of the methods described above. Effects of the invention

[0014] According to one embodiment, by mutually mapping product surface material characteristics and quality reading data to create a history database, and by automatically providing real-time feedback control of marking methods or control parameters when inspection standards are not met, it is possible to prevent the occurrence of continuous mass defects and maximize the process yield of an unmanned production line.

[0015] In addition, by converting sensing information such as illuminance, light reflectance, and surface energy obtained from the sensor unit into a multidimensional feature vector and calculating a process suitability score through matrix operations with a weighting matrix, the fundamental reliability of marking quality is significantly improved by preemptively excluding physical damage mechanisms that inevitably occur in specific surface characteristics, such as ink bleeding or laser diffuse reflection, based on hard rules.

[0016] Furthermore, by adaptively controlling the optical response of the multi-wavelength illumination module according to the characteristics of the activated marking process (ink, laser, label) and precisely extracting detailed defect measurements specialized for each marking method, such as satellite droplets, capillary bleeding, heat-affected zone discoloration, and end lifting, it is possible to establish a domain-specific precision inspection system that goes beyond simple contrast comparison.

[0017] Furthermore, it topologically divides the symbolage of the identification code into finder patterns, data encoding regions, etc., and specifically calculates the Reed-Solomon Error Correction Depletion Rate (ECC Depletion Rate) for data region defects. By assigning non-linear weights as it approaches a threshold, it quantifies even the synergistic read interference effects between multiple defects, thereby perfectly capturing the hidden risks of decoding failure.

[0018] In addition, by inputting the environmental stress transfer matrix and expected logistics transport metadata (transport time, vibration accumulation, temperature and humidity profile, etc.) into a time-series degradation prediction model (machine learning) to infer the remaining valid time (RUL) of the identification code and applying a non-linear surcharge based on the margin ratio, it predicts the probability of the barcode being damaged before the product reaches the customer, thereby providing strong business value that defends against logistics claims 100% in advance. Brief explanation of the drawing

[0019] FIGS. 1 and 2 are schematic diagrams showing a system for providing an automated marking and inspection solution for a surface printing process according to a product material according to an embodiment of the present invention. FIG. 3 is a flowchart illustrating a method for providing an automated marking and inspection solution for a surface printing process according to a product material according to an embodiment of the present invention. FIG. 4 is a flowchart showing the step of determining the marking method of a method for providing an automated solution for marking and inspection for a surface printing process according to a product material according to an embodiment of the present invention. Specific details for implementing the invention

[0020] Hereinafter, embodiments are described in detail with reference to the attached drawings. However, various modifications may be made to the embodiments, and thus the scope of the patent application is not limited or restricted by these embodiments. It should be understood that all modifications, equivalents, and substitutions to the embodiments are included within the scope of the rights.

[0021] Specific structural or functional descriptions of the embodiments are disclosed for illustrative purposes only and may be modified and implemented in various forms. Accordingly, the embodiments are not limited to the specific disclosed forms, and the scope of this specification includes modifications, equivalents, or substitutions that fall within the technical concept.

[0022] Terms such as "first" or "second" may be used to describe various components, but these terms should be interpreted solely for the purpose of distinguishing one component from another. For example, the first component may be named the second component, and similarly, the second component may be named the first component.

[0023] When it is stated that a component is "connected" to another component, it should be understood that it may be directly connected to or coupled with that other component, or that there may be other components in between.

[0024] The terms used in the embodiments are for illustrative purposes only and should not be interpreted as intended to be limiting. Singular expressions include plural expressions unless the context clearly indicates otherwise. In this specification, terms such as "comprising" or "having" are intended to indicate the existence of the features, numbers, steps, actions, components, parts, or combinations thereof described in the specification, and should be understood as not precluding the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof.

[0025] In particular, where a 'step' in this specification is described as 'comprising' one or more detailed steps or sub-steps, said 'step' may be interpreted as including its own basic processing step while simultaneously performing the described detailed steps as well.

[0026] For example, if it is stated that 'a step of doing B to A' includes 'a step of doing D to C; a step of doing F to E; and a step of doing H to G,' the 'step of doing B to A' may be interpreted not merely as the basic operation of doing B to A, but as a configuration that performs detailed procedures together, such as a step of doing D to C, a step of doing F to E, and a step of doing H to G.

[0027] Accordingly, the above configuration does not exclude various sub-procedures included within the scope of execution of the corresponding step, and may be included within the scope of the present invention even if other procedures or means performing substantially the same or equivalent functions are substituted.

[0028] Expressions such as 'end part', 'both ends', 'one end', 'other end', and 'side end' of a component can be interpreted as referring to at least / any one of the end parts of that component.

[0029] In the description of the present invention, 'a method in which a device comprises a processor, a memory, a communication module, and a non-transient storage medium, and a program stored in the non-transient storage medium is executed by the processor,' the term 'method' may be interpreted as referring to the program stored in the non-transient storage medium itself or a part of the program.

[0030] The term 'Return' as used in the description of the present invention may refer to a result value being output, returned, or returned from a method, procedure, function, etc. used in a given program language / structure.

[0031] Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as generally understood by those skilled in the art to which the embodiments pertain. Terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and should not be interpreted in an ideal or overly formal sense unless explicitly defined in this application. For example, the term 'artificial intelligence model' may be selected from one or more of known general artificial intelligence models.

[0032] In addition, when describing with reference to the attached drawings, identical components are assigned the same reference numeral regardless of drawing symbols, and redundant descriptions thereof are omitted. In describing the embodiments, if it is determined that a detailed description of related prior art could unnecessarily obscure the essence of the embodiments, such detailed description is omitted.

[0033] According to one embodiment, a method for providing an automated marking and inspection solution for a surface printing process according to a product material, wherein the device includes a processor, memory, a communication module, and a non-transient storage medium, and the processor executes a program stored in the non-transient storage medium; the method comprises: a marking method determination step of acquiring surface material data of a target product entering a production line and determining one of ink marking, laser marking, and label attachment optimized for the surface material data by referring to a pre-established marking mapping table for each material; and a marking execution step of printing or attaching an identification code on the surface of the target product by activating one of a selected mode among a first marking mode that controls the ejection of an inkjet head according to the marking method, a second marking mode that etches the surface by controlling the output and irradiation path of a laser oscillator, and a third marking mode that attaches a physical label by driving a label dispenser. An inspection execution step of acquiring an inspection image by capturing a region of interest (ROI) marked with the identification code using a vision camera, and calculating quality reading data including at least one of contrast, shape distortion, and read rate of the identification code within the inspection image based on a vision reading algorithm; and an integrated management step of mutually mapping the control parameters applied in the marking execution step with the quality reading data and storing them in a history database, and performing automatic feedback control to change the determination logic of the marking method or the control parameters corresponding to the surface material data of the target product when the quality reading data is below a preset allowable threshold.The present invention provides a method for providing an automated marking and inspection solution for a surface printing process according to product material, including: a method for determining an adaptive marking mode and automating inspection based on a multidimensional characteristic vector of the surface material; a method for inspecting optical readability quality through topological segmentation of identification codes and inference of error correction exhaustion rates; a method for predicting the degradation of identification code readability rates using a time-series environmental stress transfer matrix; a predictive maintenance marking quality management method based on logistics metadata and calculation of remaining valid time (RUL); and an automatic feedback method for printing process control parameters through multidimensional defect distribution pattern analysis.

[0034] The marking method determination step is a step of identifying the surface condition of a target product entering the production line using sensor equipment and selecting one of the first marking mode, the second marking mode, and the third marking mode based on the result.

[0035] Here, the target product may refer to containers, packaging materials, metal tags, plastic housings, films, plates, semiconductor chips, or label-attached objects that are fed into a production line via a conveyor belt, transfer robot, or automatic feeding device and require identification codes to be formed on their surfaces. Furthermore, surface material data may be a material code that directly indicates the material type of the target product, or it may be a combination of physical or chemical property data that affects marking quality, such as surface roughness, light reflectance, glossiness, surface energy, surface tension, curvature, color, transparency, thermal sensitivity, and adhesion sensitivity. Accordingly, the device can ensure the stability of subsequent marking quality by performing a material-based judgment that reflects the actual surface condition, rather than uniformly determining the marking method based solely on the product name or item name.

[0036] In the marking method determination stage, the most suitable marking method can be selected by referring together with the sensor unit, the marking mapping table by material, product identification information, and the history database.

[0037] Here, the material-specific marking mapping table may be reference information stored by linking the suitability, allowable control range, expected defect types, inspection priority items, and recommended reading criteria for ink marking, laser marking, and labeling methods, respectively, for specific surface material data based on past marking success history, experimental results, or quality data accumulated during operation. For example, in the case of paper packaging materials or matte plastics with sufficient surface energy and low reflectivity, ink marking may be considered as a priority; in the case of metals or certain coated surfaces that are hard and have good thermal reactivity, laser marking may be considered; and in cases where the application of ink or laser is unfavorable due to low surface energy, curved surfaces, or high gloss, the labeling method may be considered. As another example, if the surface material data of an incoming target product is acquired and determined to be a high-reflectivity aluminum material with a light reflectivity of 90% or more, the system may determine that there is a high risk of defects due to diffuse reflection when laser marking is applied, exclude the second marking mode, and decide one of the ink marking or labeling methods as the optimal marking method. Therefore, the marking method determination step is not a simple selection operation, but can be an adaptive judgment step that determines the most stable marking path by synthesizing material characteristics, process conditions, and past quality history.

[0038] The marking execution step is the step of actually forming an identification code on the surface of the target product according to the marking mode selected in the marking method determination step.

[0039] Here, the term "identification code" may include a data matrix, QR code, barcode, serial number, manufacturing history code, distribution tracking code, lot number, or a combination thereof, and may refer to any information marking formed on a surface so as to be optically readable. The first marking mode may be a method of directly printing a code pattern on the surface of paper, resin, or porous material by controlling the nozzle drive voltage, discharge volume, discharge frequency, spray timing, head spacing, ink viscosity, or drying conditions of an inkjet head. The second marking mode may be a method of forming an identification code by finely burning or abrading a metal or plastic surface by controlling the beam output intensity, pulse width, pulse repetition rate, focal position, and irradiation path of a scanner mirror of a laser oscillator. The third marking mode may be a method of precisely attaching a pre-printed physical label to a curved surface or a special coated material surface by driving a label dispenser, a compression roller, a pneumatic presser, and a position alignment device.

[0040] In the marking execution step, the formation mechanism may vary even for the same identification code depending on the selected mode. For example, in the case of ink marking, a cell area or linear pattern may be formed by combining multiple ink dots; in the case of laser marking, a difference in brightness may be formed by carbonization, decolorization, etching, or oxidation of the surface; and in the case of label attachment, the entire identification code printed on a separate substrate may be physically transferred to the surface of the target product. Here, control parameters are hardware setting values ​​applied to drive the marking equipment, and may include values ​​such as ink ejection pressure or nozzle pulse frequency in the first marking mode, laser peak output value or scan speed in the second marking mode, and label roller tension or compression force in the third marking mode. Therefore, the marking execution step may not be merely a step of displaying a code image, but a step of implementing a physical mark by applying control parameters corresponding to the selected material and marking mode to the actual equipment.

[0041] The inspection execution step is a step of capturing an area of ​​interest where an identification code is formed using a vision camera and quantitatively analyzing the code quality from the acquired inspection image.

[0042] Here, the region of interest may refer to a specific pixel coordinate range extracted and aligned from the entire captured image to include the symbolage of the identification code and the necessary margin area around it, and may be automatically set according to reference marks, product outlines, product reference axes, corners, or pre-registered location information. Additionally, the inspection image may include the original image captured by the vision camera or the image after lens distortion correction, luminance correction, highlight removal, specular reflection suppression, and image alignment have been performed. The vision reading algorithm may be an image processing or optical reading algorithm that receives such an inspection image as input and analyzes the cell structure, boundary sharpness, contrast, shape distortion, and actual readability of the code.

[0043] In the inspection execution step, quality read data including at least one of contrast, shape distortion, and read rate can be produced.

[0044] Here, contrast may be a numerical value representing the difference in brightness, reflective luminance, or color between the code area and the background area, and shape distortion may be a value indicating the degree of deviation from the ideal code shape due to blurring, breakage, tilting, lifting, incomplete ejection, or etching non-uniformity of the cell boundaries. Additionally, the read rate may refer to the ratio of successful attempts among multiple decoding attempts for the same identification code, or a value representing the probability of decoding according to international standards. For example, if 9 out of 10 read attempts for the same identification code are successful, the read rate may be calculated as 90%, and the shape distortion value may increase if the cell boundaries are unclear due to ink bleeding or if the code outline is distorted due to label lifting. Therefore, the inspection step may not merely perform a simple pass / fail judgment, but may also be a step for generating multidimensional quality data that can be utilized for subsequent control.

[0045] The integrated management stage is a stage that stores the control parameters actually applied during the marking execution stage and the quality reading data generated during the inspection execution stage in a linked manner, and automatically adjusts the control conditions of subsequent processes or the logic for determining the marking method based on the results.

[0046] Here, a history database may refer to a storage structure in which at least some of the following are accumulated and stored: material code, product identification information, selected marking mode, applied control parameters, inspection images, quality reading data, operation time, equipment number, and operator or line information. Accordingly, for the same material group or similar surface characteristic group, it is possible to track which control conditions in the past led to good reading rates or which conditions repeatedly caused defects; such accumulated history can be reused as reference information for determining the marking method for future products or for correcting control parameters.

[0047] In the integrated management stage, if the quality reading data is below a preset allowable threshold, the determination logic or control parameters of the marking method corresponding to the surface material data of the target product may be automatically changed. For example, if ink marking is selected but the reading rate is less than 90% and the shape distortion exceeds a reference value, the ink ejection amount may be reduced or increased, the distance between the head and the surface may be corrected, or the ink drying conditions may be adjusted. As another example, if the contrast is calculated to be low in laser marking, the laser output or scanning speed may be adjusted, and if positional deviation or edge lifting occurs repeatedly in label attachment, the label supply timing, compression force, or reference alignment offset may be modified. Furthermore, if substandard quality continues to occur in the same surface material data even after parameter correction, closed-loop control may also be performed by updating the marking mapping table by material and changing the determination logic of the marking method itself so that for products of the same material entering thereafter, the marking method is bypassed from the first marking mode to the third marking mode.

[0048] Here, automatic feedback control refers to a form of control in which the device autonomously adjusts control values ​​based on rules or learning by comparing them with reference values, even without manual interpretation by an operator of inspection results. Accordingly, the integrated management stage is not limited to a simple post-production history storage function but can be an active management stage that corrects process conditions for subsequent products within the same production batch or the next production batch in real-time or near-real-time.

[0049] Additionally, the marking method determination step comprises: calculating a surface roughness index, a light reflectance index, a glossiness index, and a surface energy estimate of a target product based on sensing information measured from a sensor unit placed on the production line, and generating a multidimensional characteristic vector including the surface roughness index, the light reflectance index, the glossiness index, and the surface energy estimate; obtaining a marking mode-specific weight matrix, an allowable threshold set, and an exclusion rule set corresponding to each of the first to third marking modes from the material-specific marking mapping table by referring to the material code, item code, required reading grade, and production speed condition included in the product identification information of the target product; and performing matrix operations on the multidimensional characteristic vector and the marking mode-specific weight matrix to calculate a mode-specific basic suitability value including ink wettability suitability for the first marking mode of the target product, laser response suitability for the second marking mode, and label adhesion suitability for the third marking mode. A step of calculating a final process suitability score for each of the first to third marking modes by reflecting, respectively, risk penalty factors including risk of ink bleeding, risk of non-adhesion, risk of laser diffuse reflection, risk of thermal damage, risk of label peeling, and risk of attachment position error for the basic suitability values ​​for each of the above modes; a step of excluding the first marking mode from the determination candidates to preemptively prevent ink drop, bleeding, or non-adhesion defects if the surface energy estimate is less than a preset threshold dyne value for securing ink wettability, or if the surface roughness index deviates from a preset roughness range that hinders uniform ink application.The method may include: a step of excluding the second marking mode from the determination candidates to prevent engraving omission, contrast degradation, or damage to surrounding optical equipment due to diffuse reflection of the laser beam when the light reflectance index or glossiness index is above a preset high-reflection threshold or the thermal sensitivity coefficient corresponding to the material code is above a preset thermal damage criterion; and a mode selection step of selecting the marking mode with the highest final process suitability score as the priority candidate method, wherein if the score difference between the priority candidate method and the next-ranked method is less than a preset competition threshold, additionally calculating a correction score reflecting past reading success rates, defect rates, and rework rates for the same material group or similar surface characteristic groups by referring to a history database, and determining the marking mode with the higher correction score as the final marking method, wherein if both the first marking mode and the second marking mode are excluded from the determination candidates, determining the third marking mode as the final marking method.

[0050] The step of calculating a surface roughness index, light reflectance index, glossiness index, and surface energy estimate of a target product based on sensing information measured from a sensor unit placed on the production line, and generating a multidimensional characteristic vector including the surface roughness index, light reflectance index, glossiness index, and surface energy estimate, is a step of converting the surface condition of a target product entering the production line into a quantitative index suitable for determining a marking method.

[0051] Here, the sensor unit may include at least one of an optical profiler for measuring curvature or fine irregularities on the surface of a target product, a spectroscopic sensor for measuring light reflection characteristics, an optical sensor for measuring glossiness, a vision camera-based surface condition estimation sensor, a capacitance-based sensor, or an auxiliary sensor for surface energy estimation. Additionally, the sensing information may be a raw measurement value directly acquired from the sensor unit, or an intermediate processing value generated by correcting, averaging, normalizing, or fusing the output values ​​of multiple sensors based on a time axis or a spatial axis. Accordingly, the device can measure the physical and chemical characteristics of the surface in real time whenever the product enters the production line and convert them into a numerical form suitable for subsequent mode selection calculations.

[0052] In the above step, a surface roughness index of the target product can be calculated. Here, the surface roughness index is a value indicating the degree of fine roughness of the surface of the target product, and may be at least one of average roughness, centerline average roughness, RMS roughness, or an estimated roughness value calculated from an optical scattering pattern. For example, if the surface roughness index is excessively low, there is a possibility that the ink may spread excessively over the surface, causing cell boundaries to smudge; conversely, if the roughness index is excessively high, the ink may penetrate unevenly between irregularities, or local response deviations may increase during laser irradiation. Additionally, the light reflectance index may be a value normalized from the ratio of reflected light to irradiated light or the distribution of reflection intensity in a specific wavelength range, and the glossiness index may be a value indicating the intensity of the specular reflection component. For example, if the light reflectance index is 88% and the glossiness index is measured as 95 GU, it may be determined to be a highly reflective and high-gloss surface; in this case, there may be a higher possibility of reading degradation due to laser diffuse reflection or highlight saturation during vision inspection. Meanwhile, the surface energy estimate may be a value indicating how stably the surface of a target product can spread or adhere liquid ink or adhesive; it may be a direct dyne measurement, or a value indirectly derived through material codes, contact angle information, surface coating information, or a pre-trained regression model. For example, if the surface energy estimate is calculated as 28 dynes, it may be interpreted as a surface unfavorable in terms of ink wettability or label adhesion stability. The device can generate a multidimensional characteristic vector with a 1×4 structure by arranging the calculated roughness index, light reflectance index, glossiness index, and surface energy estimate in a specific order, and may include additional components such as color coefficients, transparency coefficients, curvature coefficients, or contamination coefficients as needed. For example, if the surface roughness index is 0.When measured as 8 μm, light reflectance index of 88%, gloss index of 95 GU, and surface energy estimate of 28 dynes, a multidimensional characteristic vector can be generated in the form of [0.8, 88, 95, 28]. Accordingly, the device can convert different material and surface states into comparable numerical representations within the same computational structure.

[0053] The step of obtaining a marking mode-specific weight matrix, an allowable threshold set, and an exclusion rule set corresponding to each of the first to third marking modes from the material-specific marking mapping table by referring to the material code, item code, required reading grade, and production speed conditions included in the product identification information of the target product is a step of loading a judgment criterion for applying the multidimensional characteristic vector to the actual marking mode selection operation.

[0054] Here, the material code may be information for specifying basic material groups such as metal, paper, plastic, coated paper, or composite materials, and the item code may be information for distinguishing items that differ in product shape, size, purpose of use, or marking area, even if they are of the same material. Additionally, the required read grade may refer to the minimum barcode quality level or read success rate standard required in the downstream process, such as Grade A, and the production speed condition may be a condition linked to the production line throughput, conveyor transfer speed, or the allowable takt time of the marking device. Accordingly, the device can selectively call a set of computational parameters to be applied to the current product based on product-specific basic information received from the production management system or the equipment control system.

[0055] In the above step, a weight matrix for each marking mode can be obtained from the marking mapping table for each material. Here, the weight matrix for each marking mode may refer to a set of reference values ​​in the form of a matrix or table that quantifies the relative influence of each component of a multidimensional characteristic vector on the suitability of the first marking mode, the second marking mode, or the third marking mode. For example, the influence of the surface energy estimate on the suitability of the first marking mode may be set to a large positive weight, and the influence of the light reflectance index or glossiness index on the suitability of the second marking mode may be set to a negative weight. Additionally, the weight matrix may be updated according to experimental values ​​by material group, past production history, quality analysis results, or manager settings, and even for the same material code, different values ​​may be loaded depending on the item code, required reading grade, or production speed conditions. Meanwhile, the allowable threshold set may be a set of reference values ​​including the surface characteristic range in which each marking mode can be normally applied, such as the minimum surface energy value, the allowable upper limit of light reflectance, the allowable illuminance range, or the allowable thermal sensitivity range. Additionally, the set of exclusion rules may be a collection of cut-off criteria in the form of hard rules for excluding a corresponding marking mode from the candidate pool if a specific indicator satisfies a predefined risk condition. For example, it may include rules to exclude a second marking mode if the light reflectance is above a high-reflection threshold, and to exclude a first marking mode if the surface energy estimate is below a threshold value. Accordingly, the device can secure a stable judgment structure that incorporates not only simple score comparisons but also absolute exclusion conditions.

[0056] The step of performing matrix operations on the above multidimensional characteristic vector and the weight matrix for each marking mode to calculate basic suitability values ​​for each mode, including ink wettability suitability for the first marking mode of the target product, laser response suitability for the second marking mode, and label adhesion suitability for the third marking mode, is a step of primarily quantifying how well the surface characteristics of the target product match each marking method.

[0057] Here, matrix operation may be matrix multiplication, weighted summation, dot product operation, or equivalent numerical operation between a multidimensional feature vector and a weight matrix for each marking mode. For example, by performing operations on a multidimensional feature vector with a 1×4 structure and a weight matrix with a 4×3 structure, basic suitability values ​​for each mode may be calculated, such as an ink wettability suitability of 30 points for the first marking mode, a laser response suitability of 85 points for the second marking mode, and a label adhesion suitability of 75 points for the third marking mode. In this case, ink wettability suitability may be a value indicating the possibility of forming a uniform cell pattern on the surface of the target product without excessive clumping or spreading of ink; laser response suitability may be a value indicating the possibility of forming sufficient contrast by stably etching or discoloring the surface without diffuse reflection when irradiated with a laser; and label adhesion suitability may be a value indicating the possibility of the label being stably attached to the surface of the target product without wrinkles, bubbles, or edge lifting. For example, on the same production line, the highest ink wettability suitability can be calculated for matte paper boxes, the highest laser response suitability can be calculated for aluminum tags, and the highest label adhesion suitability can be calculated for high-gloss curved plastic containers.

[0058] The step of calculating a final process suitability score for each of the first to third marking modes by reflecting risk penalty factors, including the risk of ink smudging, the risk of non-adhesion, the risk of laser diffuse reflection, the risk of thermal damage, the risk of label peeling, and the risk of attachment position error, respectively, for the basic suitability values ​​for each mode, is a step of deriving a more realistic selection score by reflecting the actual possibility of defect occurrence as a subtraction factor in the suitability value calculated in the first step.

[0059] Here, the risk penalty factor may be a deduction element that quantifies the likelihood of quality degradation or process failure expected when a specific marking method is applied. For example, for the first marking mode, the risk of ink smudging and the risk of non-adhesion may be major penalty items; for the second marking mode, the risk of laser diffuse reflection and the risk of thermal damage may be major penalty items; and for the third marking mode, the risk of label peeling and the risk of attachment position error may be major penalty items. Additionally, if the risk of ink non-adhesion is calculated as 15 points under conditions of high production speed, the final process suitability score can be adjusted to 15 points by deducting 15 points from the basic suitability score of 30 points for the first marking mode. As another example, for the second marking mode, even if the basic suitability score itself is high, if the light reflectance index and the thermal sensitivity coefficient are both high, the risk of laser diffuse reflection and the risk of thermal damage increase, which can significantly lower the final process suitability score. Therefore, the above step may be a second evaluation step that determines a final score that meets actual production conditions by simultaneously considering suitability and risk.

[0060] The step of excluding the first marking mode from the determination candidates to preemptively prevent ink dropping, smudging, or non-adhesion defects when the surface energy estimate is below a preset threshold value for securing ink wettability or the surface roughness index deviates from a preset roughness range that hinders uniform ink application is a safety exclusion step that forcibly excludes ink marking regardless of the score calculation result when surface conditions exceeding the structural limit of the first marking mode are identified.

[0061] Here, the critical dyne value may be the minimum surface energy threshold value required for the ink to stably wet and spread on the surface, and the roughness range may refer to the upper and lower limit ranges of the roughness index allowed to maintain uniform cell boundaries after ink application. For example, if the estimated surface energy of the target product measured by the sensor is 28 dyne and the ink wettability threshold value defined in the allowable threshold set is 35 dyne, the surface tension is excessively low, so the ink may not stably settle on the surface and may form droplets or result in non-adherence defects with empty areas. Conversely, if the roughness index is excessively high or low, the ink film thickness becomes uneven, causing the boundaries of the code cells to smudge or become distorted. Therefore, the device can reduce the discrepancy between the calculated score and actual process stability and prevent the selection of an incorrect method by completely removing the first marking mode from the candidate pool rather than merely deducting points.

[0062] The step of excluding the second marking mode from the determination candidates to prevent omission of engraving, reduction of contrast, or damage to surrounding optical equipment due to diffuse reflection of the laser beam when the light reflectance index or glossiness index is above a preset high-reflection threshold or the thermal sensitivity coefficient corresponding to the material code is above a preset thermal damage standard is a step of forcibly excluding the second marking mode when the risk of application of laser marking exceeds a standard value.

[0063] Here, the high-reflectivity threshold may be a reference value at which the laser beam or inspection light source is excessively reflected from the surface, increasing the likelihood of interfering with stable etching or optical reading. Additionally, the thermal sensitivity coefficient is a material-specific thermal vulnerability parameter stored in correspondence with the aforementioned material code, representing a value indicating the likelihood of surface discoloration, melting, warping, or internal damage occurring during laser irradiation. For example, if the light reflectance index is measured at 88%—such as in an aluminum panel—exceeding the high-reflectivity threshold of 80%, the high-power laser beam may fail to etch the surface stably and instead undergo specular reflection, potentially resulting in missing engravings or reduced contrast. Furthermore, in the case of thin polyethylene films with high thermal sensitivity coefficients or composite packaging materials with heat-vulnerable coating layers, the surface may melt or shrink due to laser heat, potentially worsening reading quality. Moreover, diffuse reflection on a high-reflectivity surface can cause not only quality issues with the target product itself but also problems regarding equipment safety by allowing unnecessary light energy to reach surrounding cameras, sensors, or protective covers. Therefore, the device can preemptively block structural risks associated with laser application by excluding the second marking mode from the candidate group.

[0064] The mode selection step, which selects the marking mode with the highest final process suitability score as the priority candidate method, and if the score difference between the priority candidate method and the next-ranked method is less than a preset competition threshold, additionally calculates a correction score reflecting the past reading success rate, defect rate, and rework rate for the same material group or similar surface characteristic group by referring to a history database, and determines the marking mode with the higher correction score as the final marking method, and if both the first marking mode and the second marking mode are excluded from the determination candidates, determines the third marking mode as the final marking method, is a step of determining the final application mode by synthesizing the preceding suitability calculation results and exclusion judgment results.

[0065] Here, the first candidate method may refer to the method with the highest final process suitability score among the remaining marking modes that were not excluded, and the second-ranked method may refer to the method with the next highest score. Additionally, the competition threshold may be a reference value for determining a state where it is difficult to make a stable decision based solely on a simple numerical comparison, as the difference between the two scores is very small. For example, if the final process suitability score of the first marking mode is 75 points and the final process suitability score of the third marking mode is 74 points, and the difference between the two is only 1 point, which is less than the competition threshold of 5 points, a correction judgment reflecting past history may be performed instead of simply selecting the highest score.

[0066] The above step may additionally calculate a correction score by referring to the history database. Here, the correction score may be a value that adjusts the effectiveness of the current candidate method by reflecting factors actually achieved in the past within the same material group or similar surface characteristic group, such as reading success rates, recurring defect rates, rework frequency, equipment stoppage counts, or quality retention rates in subsequent logistics processes. For example, even if the current scores of the first marking mode and the third marking mode are similar, if the history shows that the first marking mode had a low reading success rate and a high rework rate within the same material group, while the third marking mode had a high final yield rate despite higher initial costs, the correction score for the third marking mode may be calculated higher, allowing it to be selected as the final method. Furthermore, if both the first and second marking modes are excluded from the candidate group, the third marking mode may be immediately determined as the final marking method as a safety mode or bypass mode. Accordingly, the mode selection step may be a final judgment step for selecting a method capable of providing the most stable marking results in the actual production site by reflecting not only current measurement values ​​but also past quality history.

[0067] And, the inspection execution step comprises: a step of controlling the irradiation angle, irradiation direction, visible light wavelength band, polarization state, and light intensity of a multi-wavelength illumination module based on an activated marking mode, expected marking location information of a target product, and optical response characteristics of the surface of a target product, and acquiring a diffuse reflection inspection image in which specular reflection components are suppressed; a step of performing lens distortion correction, perspective correction, luminance flattening, highlight removal, and image alignment on the diffuse reflection inspection image, and generating an aligned ROI image by extracting a region of interest marked with an identification code by referencing at least one of the outer contour, reference mark, reference hole, corner, and product reference axis of the target product; and a step of calling an image processing recipe pre-mapped to the activated marking mode and selectively applying at least one morphological filter among an adaptive binarization algorithm, a contrast enhancement filter, an edge enhancement filter, and dilation, erosion, opening, and closing to the aligned ROI image, thereby segmenting the cell area, code boundary area, and background area of ​​the identification code. When the activated marking mode is a first marking mode, the step of measuring the distribution density of satellite droplets, the capillary spreading area between cells, the ratio of unejected blank pixels, the deviation in ink diffusion directionality, and the boundary roughness index of the code outer edge based on the cell area and the code boundary area, and deriving a first mode index including an ink spreading defect index and a shape distortion index using the distribution density of satellite droplets, the capillary spreading area between cells, the ratio of unejected blank pixels, the deviation in ink diffusion directionality, and the boundary roughness index of the code outer edge;When the above-mentioned activated marking mode is a second marking mode, the method comprises the step of measuring the discoloration spectrum deviation of the heat-affected zone, the frequency of occurrence of microcrack candidate patterns at the etching interface, the optical response non-uniformity at the laser beam overlap point, and the contrast deviation corresponding to the carbonized edge width and etching depth using the optical response difference due to multi-angle illumination and the change in reflected luminance by wavelength band, and deriving a second mode index including a laser etching integrity index and a contrast index using the discoloration spectrum deviation of the heat-affected zone, the frequency of occurrence of microcrack candidate patterns at the etching interface, the optical response non-uniformity at the laser beam overlap point, and the contrast deviation corresponding to the carbonized edge width and etching depth; If the activated marking mode is a third marking mode, the method comprises the steps of: performing contour tracking and edge profile analysis along with code outline information required for reading the identification code to measure an end lifting index based on changes in the radius of curvature of the label edge, a label adhesion misalignment angle relative to the product reference axis, an area of ​​bubble occurrence within the adhesive surface, a wrinkle distribution density, and a positional deviation of the label outline; deriving a third mode index including a physical adhesion integrity index using the end lifting index, the label adhesion misalignment angle relative to the product reference axis, an area of ​​bubble occurrence within the adhesive surface, a wrinkle distribution density, and a positional deviation of the label outline; calculating a basic read rate value of the cell area of ​​the identification code based on international standard barcode specifications; generating a defect feature vector by normalizing the detailed defect measurement value derived corresponding to the activated marking mode and the mode index corresponding to the activated marking mode; and calculating a weighted penalty based on the severity of each item included in the defect feature vector. A step of calculating an effective reading rate by reflecting a point deduction weighting penalty to the above-mentioned basic reading rate value;The method may include: a step of analyzing a multidimensional defect distribution pattern from the above defect feature vector to calculate a degradation prediction score that quantifies physical vulnerability to friction, vibration, and temperature and humidity change environments in the logistics post-process; a step of calculating an initial read failure risk based on the difference between the above valid read rate and a preset target read threshold value, and calculating a readability conversion risk by reflecting the degradation prediction score in the initial read failure risk; and a step of generating final quality read data by mapping the above valid read rate, defect feature vector, degradation prediction score, and readability conversion risk.

[0068] The inspection execution step is a step of reliably capturing an area where an identification code is formed based on an activated marking mode, expected marking location information of the target product, and optical response characteristics of the target product surface, and quantitatively interpreting defect patterns that appear differently depending on the marking method to generate final quality reading data.

[0069] Here, the predicted marking location information may refer to area information where an identification code is predicted to exist, based on the target product's reference shape, product dimensions, installation coordinates of the marking head or label applicator, transport position on the conveyor, product reference axis, and a pre-set code formation position. Additionally, optical response characteristics may refer to surface-specific optical response characteristics, including reflectance, glossiness, polarization dependence, color response, diffuse reflection distribution, or the intensity of specular reflection components, which the target product surface exhibits in response to illumination light. Accordingly, the device does not uniformly inspect all products with a fixed camera setting; instead, it first establishes inspection conditions that adaptively respond to the currently active marking mode and surface condition, and then acquires images optimized for those conditions.

[0070] Based on the activated marking mode, expected marking location information of the target product, and optical response characteristics of the target product surface, the step of controlling the irradiation angle, irradiation direction, visible light wavelength band, polarization state, and light intensity of the multi-wavelength illumination module, and acquiring a diffuse reflection inspection image with suppressed specular reflection components, is a step of securing an original image suitable for subsequent image processing and reading operations.

[0071] Here, the multi-wavelength illumination module may be a set of light sources capable of selectively illuminating multiple visible light wavelength bands, and the polarization state can be adjusted through the arrangement of polarization filters or the control of the polarization direction. For example, for surfaces that generate strong glare, such as metal panels or vinyl-coated materials, by selectively illuminating with light of a specific wavelength band or changing the polarization state, specular reflection light entering directly into the camera lens can be suppressed, and a diffuse reflection inspection image can be obtained in which only the shape of the identification code is clearly revealed. Additionally, when the first marking mode is activated, the illumination angle and light intensity can be adjusted to maximize the contrast difference between the ink dot and the background; when the second marking mode is activated, light of a specific wavelength band can be applied preferentially to clearly reveal the difference in reflective luminance between the etched and non-etched parts; and when the third marking mode is activated, the illumination direction and polarization state can be adjusted to clearly reveal local reflection distortion caused by lifting or bubbles on the label surface. Accordingly, the device can obtain input images that enable subsequent analysis even for target products sensitive to specular reflection, such as high-gloss surfaces, metal surfaces, or label-coated surfaces.

[0072] The step of generating an aligned ROI image by performing lens distortion correction, perspective correction, luminance flattening, highlight removal, and image alignment on the diffuse reflection inspection image, and extracting a region of interest marked with an identification code by referencing at least one of the outer contour, reference mark, reference hole, corner, and product reference axis of the target product, is a step of separating only the valid area required for identification code analysis from the acquired inspection image in a normalized form.

[0073] Here, lens distortion correction may be an operation that compensates for barrel distortion or pincushion distortion of the camera lens, and perspective correction may be an operation that restores the code area, which appears trapezoidal due to differences in shooting angles, to a planar state aligned with horizontal and vertical references. Additionally, luminance flattening may be a process that corrects brightness imbalances caused by deviations in lighting distribution, and highlight removal may be a process that suppresses local saturation areas to ensure that code cells or boundaries are stably distinguished from the background. Here, the ROI may be an image precisely extracted and aligned from a wide product surface image to include the symbolism of the identification code and the Quiet Zone margin required by standards. Therefore, the device can consistently interpret the code structure in a reference coordinate system in subsequent steps.

[0074] The step of segmenting the cell region, code boundary region, and background region of the identification code by calling an image processing recipe pre-mapped to the activated marking mode and selectively applying at least one morphological filter among an adaptive binarization algorithm, a contrast enhancement filter, an edge enhancement filter, and dilation, erosion, opening, and closing to the aligned ROI image is a preprocessing and segmentation step for reliably detecting different defect characteristics for each mode.

[0075] Here, an image processing recipe may refer to a set of pre-set image processing parameters based on the marking mode, material group, code type, inspection lighting conditions, and past quality history. For example, for the first marking mode, adaptive binarization and closing operations may be applied first to emphasize ink bleeding and unextruded gaps; for the second marking mode, edge enhancement filters and contrast enhancement filters may be applied first to increase the contrast of etching boundaries and carbonized edges; and for the third marking mode, opening operations and erosion operations may be applied selectively to clarify label outlines and lifted areas. Additionally, by removing unnecessary fine noise through a morphology filter, the cell area, which is the actual marking area constituting the identification code, the code boundary area where the cell meets the background, and the non-marking background area can be clearly separated. Accordingly, in the subsequent defect measurement step, the device can specifically determine whether each defect is a problem with the cell itself, a problem with the boundary line, or a problem with distinguishing it from the background.

[0076] When the activated marking mode is a first marking mode, the step of measuring the distribution density of satellite droplets, the capillary bleeding area between cells, the ratio of unejected blank pixels, the deviation in ink diffusion directionality, and the boundary roughness index of the code outer edge based on the cell area and the code boundary area, and deriving a first mode index including an ink spreading defect index and a shape distortion index using the distribution density of satellite droplets, the capillary bleeding area between cells, the ratio of unejected blank pixels, the deviation in ink diffusion directionality, and the boundary roughness index of the code outer edge is a step of quantitatively quantifying printing defects that occur uniquely in ink-based marking.

[0077] Here, satellite droplets may refer to fine ink particles formed by splashing out to the periphery in addition to the intended main ink droplet, and the distribution density of satellite droplets may be a value representing the number or degree of concentration of these fine particles per unit area. Additionally, the inter-cell capillary bleeding area may refer to the size of the region where cells that should be separated become entangled as ink seeps between adjacent cells along the grain of paper or porous material, and the unejected blank pixel ratio may be the ratio of pixels that are not actually coated within the area where ink should normally be applied. The ink diffusion directionality deviation may be a value representing the degree to which the ink spreads in a biased direction, and the boundary roughness index of the code outline may be a value representing the degree to which the outline deviates from an ideal straight line or curve. The device can derive an ink spreading defect index and a shape distortion index by synthesizing these detailed defect measurements, thereby estimating causes such as excessive ink wettability, nozzle condition abnormalities, surface absorption deviations, or printing pressure imbalances.

[0078] When the above-mentioned activated marking mode is a second marking mode, the step of measuring the discoloration spectrum deviation of the heat-affected zone, the frequency of occurrence of candidate microcrack patterns at the etching interface, the optical response non-uniformity at the laser beam overlap point, and the contrast deviation corresponding to the carbonization edge width and etching depth using the optical response difference due to multi-angle illumination and the change in reflected luminance by wavelength band, and deriving a second mode index including a laser etching integrity index and a contrast index using the discoloration spectrum deviation of the heat-affected zone, the frequency of occurrence of candidate microcrack patterns at the etching interface, the optical response non-uniformity at the laser beam overlap point, and the contrast deviation corresponding to the carbonization edge width and etching depth, is a step of analyzing surface damage and etching non-uniformity specialized for the laser engraving method.

[0079] Here, the heat-affected zone (HAZ) may refer to an area around the point where the laser beam is directly irradiated where high heat is conducted, causing the material's properties or color to change. The discoloration spectrum deviation may be a value indicating the degree to which the reflection characteristics of the HAZ by wavelength range deviate from a reference state. Additionally, the frequency of microcrack candidate patterns may be a value indicating how many linear patterns presumed to be cracks occur per unit length or unit area around the etching interface or the HAZ. The optical response non-uniformity may refer to the degree to which cells or line segments that should be formed under the same etching conditions actually exhibit different reflection luminance or color responses. The carbonization edge width may be the width of the carbonized or discolored area formed around the laser irradiation boundary. The contrast deviation corresponding to the etching depth may refer to the degree to which a difference in etching depth appears as a difference in contrast in the optical image. The device can derive laser etching integrity and contrast indices by synthesizing these detailed defect measurements, thereby determining the possibility of excessive or insufficient laser output, irradiation path overlap, material response deviation, or thermal damage.

[0080] When the above-mentioned activated marking mode is a third marking mode, the step of performing contour tracking and edge profile analysis along with code outline information required for reading the identification code to measure the end lifting index based on the change in the radius of curvature of the label edge, the label adhesion misalignment angle relative to the product reference axis, the bubble generation area within the adhesive surface, the wrinkle distribution density, and the positional deviation of the label outline, and deriving a third mode index including a physical adhesion integrity index using the end lifting index, the label adhesion misalignment angle relative to the product reference axis, the bubble generation area within the adhesive surface, the wrinkle distribution density, and the positional deviation of the label outline is a step of analyzing whether there is an abnormality in the adhesive state specialized for the physical label attachment method.

[0081] Here, code outline information may refer to outline information including the outer boundary required for reading the identification code itself, the label boundary, and the dividing line between the code and the background. Additionally, the end lifting index may be a numerical value indicating the degree to which the label edge is separated from the surface based on changes in the radius of curvature of the label edge or local shadow patterns, and the label adhesion misalignment angle relative to the product reference axis may be an angle value indicating the degree to which the label is attached with a rotational error relative to the product reference axis. The bubble area within the adhesive surface may be the total area of ​​the non-attached region formed by trapped air within the label adhesive surface, the wrinkle distribution density may refer to the degree of distribution per unit area of ​​the wrinkle structure formed on the label surface, and the label outline positional deviation may be the distance value from the target attachment position where the actual outline deviates. The device can derive a physical attachment integrity index using these detailed defect measurements, thereby estimating causes such as label tension imbalance, insufficient compression force, unsuitable surface curvature, or alignment errors. In addition, the above physical attachment integrity indicator can be used as an indicator of the possibility that the label may be torn off or damaged by external friction during future logistics transportation.

[0082] The step of calculating the basic read rate of the cell area of ​​the identification code based on international standard barcode specifications, generating a defect feature vector by normalizing the detailed defect measurement derived corresponding to the activated marking mode and the mode indicator corresponding to the activated marking mode, and calculating a weighted penalty based on the severity of each item included in the defect feature vector is a step of organizing defects by mode into standardized quality evaluation input values.

[0083] Here, the basic reading rate may be a basic reading score in the form of grades A to F or % derived when applying international standard decoding algorithms such as ISO. Additionally, the defect feature vector may be a structure in which detailed defect measurements having different measurement units and scales, such as area, angle, density, or deviation values, are normalized to the same scale, such as a range of 0 to 1, and grouped into a single one-dimensional mathematical array data. For example, the first marking mode may include the distribution density of satellite droplets, the capillary spreading area, the ratio of undischarged blank pixels, and the first mode indicator; the second marking mode may include the discoloration spectrum deviation, the frequency of occurrence of microcrack candidate patterns, and the second mode indicator; and the third marking mode may include the end lifting index, the bubble occurrence area, and the third mode indicator. The above-mentioned deduction weight penalty may be a penalty value calculated by reflecting the severity with which a specific defect element within the defect feature vector affects read performance degradation, and a larger deduction weight may be assigned to defects that directly affect reading, such as cell structure damage, boundary collapse, or direct contrast degradation. Accordingly, the device can organize the severity of each defect in a comparable form, regardless of which marking mode the current inspection target was formed by.

[0084] The step of calculating the effective read rate by reflecting a weighted penalty to the above-mentioned basic read rate value is a step of deriving a practical read performance value that considers both the standard decoding results and the actual defect severity. For example, even if an excellent basic read rate of 95 points is calculated based on general standard inspection criteria, if a significant weighted penalty is calculated due to defects such as micro-etching cracks, boundary collapse, end lifting, or unextruded gaps measured by this system, the effective read rate reflecting this may be adjusted downward. Conversely, even if the basic read rate value is somewhat low, the extent of the deduction may be limited if defects are confined to the periphery and the core cell structure is stable. Therefore, the effective read rate is not merely a value reflecting the success of decoding at the current point in time, but may represent a practical quality level that considers the current defect status and the possibility of durability degradation.

[0085] The step of analyzing the multidimensional defect distribution pattern from the above defect feature vector to calculate a degradation prediction score that quantifies physical vulnerability to friction, vibration, and temperature and humidity change environments in the post-logistics process is a step of predicting how quickly the currently detected defect can lead to quality degradation in the logistics environment after product shipment.

[0086] Here, the multidimensional defect distribution pattern may be pattern information derived by analyzing the magnitude, combination, concentration, and intercorrelation of items included in the defect feature vector. For example, if the end lifting index and bubble area appear significantly simultaneously in the third marking mode, the likelihood of rapid label peeling due to repeated friction or changes in temperature and humidity may increase; similarly, if the microcrack candidate pattern and contrast deviation corresponding to etching depth appear significantly simultaneously in the second marking mode, the degradation of code contrast may accelerate due to vibration or thermal changes. Furthermore, if the material is exposed to a high-temperature and high-humidity logistics environment for a long period while ink bleeding components are already present in the first marking mode, readability may further deteriorate due to additional ink diffusion. Therefore, the degradation prediction score may serve as a predictive indicator indicating the potential for future readability degradation beyond the current pass / fail status.

[0087] The step of calculating an initial read failure risk based on the difference between the above-mentioned effective read rate and a preset target read threshold value, and calculating a readability conversion risk by reflecting a degradation prediction score in the above-mentioned initial read failure risk, is a comprehensive risk calculation step that considers both the current quality level and the possibility of future degradation.

[0088] Here, the target reading threshold refers to the minimum effective reading rate threshold that must be secured for product shipment or input into subsequent processes. For example, if the effective reading rate is sufficiently higher than the target reading threshold, the risk of initial reading failure may be calculated as low; however, if the degradation prediction score is high, an additional risk of transitioning to an unreadable state during subsequent logistics processes may be reflected. Conversely, if the current effective reading rate is close to or already low than the threshold, and the degradation prediction score is also high, the risk of transitioning to an unreadable state may increase significantly. Therefore, the above step can serve to quantitatively reveal potential defects that are not apparent from current inspection results alone.

[0089] The step of generating final quality read data by mapping the above-mentioned valid read rate, defect feature vector, degradation prediction score, and unreadable conversion risk is a step of organizing all preceding inspection and analysis results into an integrated data structure suitable for subsequent control and history management.

[0090] Here, the final quality reading data may include not only whether a simple reading was successful, but also an effective reading rate representing the actual reading performance at the current point in time, a defect feature vector structurally expressing the type and severity of defects, a degradation prediction score indicating vulnerability to subsequent logistics environments, and a risk of transition to an unreadable state indicating the likelihood of transitioning to an unreadable state. Furthermore, the generated final quality reading data is stored in the control unit memory or history DB and then transmitted to the integrated management stage, where it can be utilized as reference information to inversely correct the control parameters of the printing equipment, update the marking method determination logic for the same material group, or determine whether a specific production batch requires rework. Accordingly, the inspection execution stage does not remain at the level of simple imaging and reading, but can be an intelligent inspection stage that includes precise analysis of defects by marking method, current quality judgment, future degradation prediction, and subsequent control linkage.

[0091] In addition, the step of calculating the above-mentioned weighted penalty comprises: aligning a symbolage layout corresponding to the specification information of the identification code onto a reference coordinate system, and then spatially dividing the symbolage layout to establish a plurality of topological importance zones including a finder pattern area, a timing pattern area, a data encoding area, and a quiet zone area, wherein if at least one of an alignment pattern area, a version information area, and a format information area exists according to the identification code specification information, this is set as an additional auxiliary importance zone; and, based on the spatial data of the aligned ROI image and the divided cell area and code boundary area, mapping the pixel coordinates of detailed defect measurements, defect area, defect outline shape, and defect clustering diagram to the reference coordinate system to generate a zone-specific defect distribution matrix including the defect occupancy area, defect occupancy ratio, defect clustering diagram, defect center coordinate distribution, and continuous intrusion length of defects for each of the topological importance zones. A step of identifying an error correction level corresponding to the international standard barcode specification of the identification code, and calculating an error correction exhaustion rate representing the ratio of codewords damaged due to defects relative to the total error correction restoration capacity of the identification code, based on the defect occupancy ratio and the continuous intrusion length of defects within the data encoding area; a step of, for the zone-specific defect distribution matrix, assigning a first weight of the highest grade to defects occurring in the finder pattern area and timing pattern area, reflecting the risk of direct connection to unrecognizability due to loss of decoding reference axis or code alignment failure; assigning a second weight to defects occurring in the quiet zone area, reflecting the level of background separation margin infringement; and assigning an auxiliary weight to defects occurring in the alignment pattern area, version information area, or format information area, reflecting the level of code consistency degradation.For defects occurring in the data encoding area, a step of calculating a third weight by inputting the error correction exhaustion rate, defect clustering, and continuous intrusion length of the defect into a weighting function defined such that the weight increases non-linearly based on a threshold as the error correction exhaustion rate approaches a preset limit restoration rate; a step of calculating a zone-specific penalty component value by combining the corresponding weight, defect occupancy area, and defect occupancy ratio for each of the finder pattern area, timing pattern area, data encoding area, quiet zone area, and auxiliary importance area, and calculating a first penalty value by weighted summing the zone-specific penalty component values; The method may include the step of calculating a synergy coefficient that reflects the reading interference effect due to mutual interference between multiple defects when, within the defect feature vector, at least two detailed defect measurements among the ratio of undischarged blank pixels, the area of ​​inter-cell capillary bleeding, the contrast deviation corresponding to the etching depth, the frequency of occurrence of microcrack candidate patterns at the etching interface, the end lifting index, and the label adhesion misalignment angle relative to the product reference axis overlap across adjacent topological importance zones, or when the clustering of detailed defect measurements that directly impair the contrast of the code exceeds a preset threshold; and the step of deriving a final deduction weighted penalty by reflecting the synergy coefficient in the first penalty value, wherein an additional upper limit coefficient is applied so that the penalty value before and after the application of the synergy coefficient does not exceed a preset stability limit.

[0092] The step of calculating the above-mentioned weighted penalty is a step of precisely calculating the final amount of deduction by not uniformly evaluating defects occurring in the identification code based on a simple defective area, but by reflecting together where each defect exists within the code in terms of its functional location, the extent to which damage at that location directly contributes to actual read failure, and how much multiple defects amplify the read interference effect through mutual interference.

[0093] Here, the deduction weight penalty may be a quantitative value in the nature of a penalty that is directly reflected in the basic read rate to lower the effective read rate, and can be utilized as a key input value to calculate the risk of conversion to unreadable status and the final quality read data in subsequent stages. Therefore, the above stage does not remain merely a simple defect detection stage but may be a structural quality evaluation stage that comprehensively reflects code structure, defect location, error correction margin, and the mutual interference effects between complex defects.

[0094] After aligning the symbolage layout corresponding to the specification information of the identification code above on a reference coordinate system, the symbolage layout is spatially divided to set a plurality of topological importance zones including a finder pattern area, a timing pattern area, a data encoding area, and a quiet zone area, wherein if at least one of the alignment pattern area, version information area, and format information area exists according to the identification code specification information, it is set as an additional auxiliary importance zone, which is a step of reconstructing the geometric structure of the code to be inspected into functional risk zones.

[0095] Here, the symbolage layout may refer to the entire code structure in which cell arrangements, reference patterns, data storage areas, alignment aid areas, and margin configurations are defined according to Data Matrix, QR code, or other 2D code standards. Additionally, the topological importance area may be a zone classified based on read criticality according to the logical role performed by each cell or pattern. For example, the finder pattern area may be a reference area for the reader to determine the overall location and orientation of the code, the timing pattern area may be a reference area for interpreting cell spacing and alignment consistency, and the data encoding area may be a core area where actual product information is encrypted and stored. Furthermore, the quiet zone area may be an outer margin required to separate background noise from the code area, and the alignment pattern area, version information area, and format information area may be auxiliary function areas to maintain the accuracy of code alignment and standard interpretation. Therefore, even if the same defect is present, the device may interpret it with different read risk levels depending on which functional area it is located in.

[0096] The step of mapping the pixel coordinates of detailed defect measurements, defect area, defect outline, and defect clustering diagram to the reference coordinate system based on the spatial data of the aligned ROI image and the segmented cell area and code boundary area, and generating a zone-specific defect distribution matrix including the defect occupancy area, defect occupancy ratio, defect clustering diagram, defect center coordinate distribution, and continuous intrusion length of defects for each of the topological importance zones, is a quantitative mapping step that corresponds physical defects detected through vision inspection to a code structure.

[0097] Here, Spatial Data may be data containing coordinate and shape information indicating the locations of cell regions, code boundary regions, and background regions on an aligned ROI image. Additionally, the defect distribution matrix by region may refer to a two-dimensional or multi-dimensional matrix structure that records defect occupancy area, defect occupancy ratio, defect clustering, defect center coordinate distribution, and continuous encroachment length in numerical form for each importance region. For example, if a specific defect is locally densely concentrated within a data encoding region, a high defect clustering ratio may be calculated, and if the same defect extends for a long distance in the cell row or column direction, a large continuous encroachment length may be calculated. Therefore, the device can not only record the presence or absence of defects but also represent, in the form of a structured matrix, the proportion at which defects are distributed within a functional region, what shape and density characteristics they possess, and how continuously they damage the code structure.

[0098] The step of identifying an error correction level corresponding to the international standard barcode specification of the identification code, and calculating an error correction exhaustion rate representing the ratio of codewords damaged by defects relative to the total error correction restoration capacity of the identification code based on the defect occupancy ratio and the continuous intrusion length of defects within the data encoding area, is a step of quantitatively evaluating how much of the error restoration margin allowed by the code specification is consumed by data area damage.

[0099] Here, the error correction level may refer to the level of error recovery capability embedded in the QR code or Data Matrix specifications, and a codeword may be the smallest unit data block used for data and error correction. Additionally, the error correction exhaustion rate may be a value representing, as a percentage or ratio, how much of the total recoverable capacity has already been used due to the current defect. For example, in a code designed to recover up to 20 damaged codewords in total, if it is determined that 15 codewords are substantially damaged due to the current defect, the error correction exhaustion rate may be calculated at the level of 75%. Furthermore, even with the same defect occupancy area, if the defect encroaches continuously along a specific row or column for a long distance rather than being scattered, the likelihood of multiple codewords being damaged simultaneously increases, so the error correction exhaustion rate may be calculated to be higher. Accordingly, the device can interpret data encoding area damage more precisely from the perspective of actual error correction margin consumption rather than based on a simple area criterion.

[0100] Regarding the defect distribution matrix by zone above, the step of assigning a first weight of the highest grade reflecting the risk of direct connection to unrecognizability due to loss of decoding reference axis or code alignment failure to defects occurring in the finder pattern area and timing pattern area, assigning a second weight reflecting the level of background separation margin infringement to defects occurring in the quiet zone area, and assigning an auxiliary weight reflecting the level of code consistency degradation to defects occurring in the alignment pattern area, version information area, or format information area is a step of differentially setting penalty criteria according to the functional importance of the area where the defect is located.

[0101] Here, the first weight may be set to the highest level, reflecting that damage to the finder pattern or timing pattern can lead directly to failure in determining code orientation, cell alignment, or grid reference setting. The second weight may be set to a level corresponding to the possibility of background separation failure, reflecting that the quiet zone is an essential margin for separating the background from the code. Additionally, since the alignment pattern area, version information area, or format information area is related to code alignment assistance and specification interpretation accuracy, an auxiliary weight may be assigned that reflects significance, although lower than that of the finder pattern. Therefore, the device may not treat cases occurring in the finder pattern area and cases occurring in the quiet zone area in the same way, even if the defect area is the same, but may assign different penalty contributions by reflecting the readability of each functional area.

[0102] Regarding defects occurring in the above data encoding area, the step of calculating a third weight by inputting the error correction exhaustion rate, defect clustering, and the continuous intrusion length of the defect into a weighting function defined such that the weight increases non-linearly based on a threshold as the error correction exhaustion rate approaches a preset limit restoration rate is a step that dynamically reflects the risk that data storage area damage will turn into actual decoding failure.

[0103] Here, the limit recovery rate may be the upper limit level at which the error correction function can reliably recover according to the code specification, and the threshold may be the reference value of the transition section where the error correction margin begins to be rapidly exhausted. Additionally, the defect clustering and the continuous intrusion length of defects may be correction factors that reflect whether defects are dispersed or whether they are concentrated in a specific section and continuously damage the data area, even if the occupancy ratio is the same. For example, the third weight may increase gradually in sections where the error correction exhaustion rate is low, but by setting the weight to rise sharply as the exhaustion rate approaches the limit recovery rate, a cliff effect can be reflected where normal reading is maintained until just before reaching the error correction limit, and then plummets to an unreadable state the moment the limit is exceeded. Accordingly, the device can evaluate data encoding area damage not as a static linear deduction, but as a non-linear risk based on the actual mathematical recovery limit.

[0104] For each of the above-mentioned finder pattern area, timing pattern area, data encoding area, quiet zone area, and auxiliary importance area, the step of calculating a penalty component value for each area by combining the corresponding weight, the defect occupancy area, and the defect occupancy ratio for each area, and calculating a first penalty value by weighted summing the penalty component values ​​for each area, is a step of primarily integrating the deduction factors identified across the entire code area.

[0105] Here, the zone-specific penalty component value may be a partial penalty calculated by combining a defect occurring in each functional zone with the importance weight of that zone. For example, even with the same defect occupancy rate, if it exists in the finder pattern area, a first weight may be applied to calculate a large penalty component value, and if it exists in the quiet zone area, a second weight may be applied to calculate a relatively small value. Additionally, for the data encoding area, a third weight may be applied to calculate a penalty component value that reflects the error correction exhaustion state and defect density characteristics. The device may derive a first penalty value by weighted summing the multiple zone-specific penalty component values ​​calculated in this way, and the first penalty value may be a comprehensive penalty that summarizes the degree of code structure damage from the perspective of importance by functional zone.

[0106] In the above defect feature vector, if at least two detailed defect measurements among the unejected blank pixel ratio, inter-cell capillary spreading area, contrast deviation corresponding to etching depth, frequency of occurrence of microcrack candidate patterns at the etching interface, end lifting index, and label adhesion skew angle relative to the product reference axis overlap across adjacent topological importance zones, or if the clustering of detailed defect measurements that directly impair the contrast of the code exceeds a preset threshold, the step of calculating a synergy coefficient that reflects the reading interference effect due to mutual interference between multiple defects is a step that separately reflects a situation in which multiple defect factors combine to cause a recognition failure greater than simple summation. For example, in the first marking mode, if the unejected blank pixel ratio and the inter-cell capillary spreading area occur simultaneously near the boundary between the quiet zone and the data encoding area, cell separation failure and boundary damage act in combination, which may amplify the confusion experienced by the reader. Furthermore, in the second marking mode, if the contrast deviation corresponding to the etching depth and the frequency of occurrence of candidate microcrack patterns at the etching interface appear significantly simultaneously, structural discontinuity may be intensified beyond simple contrast degradation. In the third marking mode, if the end lifting index and the label adhesion misalignment angle relative to the product reference axis are significantly high simultaneously, the overall positional stability of the code and the outer boundary stability may be simultaneously degraded. Therefore, the device can reflect the mutual interference effect of complex defects as a separate amplification factor by calculating a synergy coefficient based on whether multiple defects overlap across adjacent importance zones or whether defects that directly impair contrast exhibit a high degree of clustering.

[0107] The step of deriving a final penalty with weighted deductions by reflecting a synergy coefficient in the first penalty value above, while additionally applying an upper limit coefficient so that the penalty value before and after the application of the synergy coefficient does not exceed a pre-set stability limit, is a step of determining the final penalty that reflects both the importance of each zone and the mutual interference effect between complex defects.

[0108] Here, the synergy factor may be a value indicating the extent to which the read interference effect expands beyond simple addition when multiple defects exist simultaneously, and the upper limit factor may be a stabilization factor to prevent the result of applying the synergy factor from diverging into an abnormally excessive penalty. For example, even if the primary penalty value itself is at an intermediate level, if damage to the finder pattern area and violation of the quiet zone occur simultaneously and the defect clustering degree of the data encoding area is also high, the synergy factor may be applied to increase the final weighted penalty. However, if the result of the increase exceeds a pre-set stability limit, the upper limit factor may be applied to limit the penalty upper limit. Accordingly, the weighted penalty may be utilized as a structural quality degradation indicator that reflects not only the simple defect area ratio or the simple deduction of read scores, but also the code specification structure, error correction margin, importance by functional area, and the mutual interference effects between complex defects.

[0109] Additionally, the step of calculating the degradation prediction score comprises: extracting a multidimensional defect distribution pattern representing the damage location, damage size, damage density, and damage propagation directionality of the identification code based on the defect feature vector, the defect cluster diagram included in the zone-specific defect distribution matrix, the defect occupancy ratio, and the continuous intrusion length of the defect, the valid read rate, and the deduction weighted penalty; projecting the multidimensional defect distribution pattern onto a pre-established environmental stress transfer matrix to calculate the basic sensitivity for each environmental factor for the friction factor, vibration factor, temperature change factor, humidity change factor, repetitive impact factor, and repetitive contact factor applied to the identification code; and, when the activated marking mode is the first marking mode, up-correcting the basic sensitivity corresponding to the temperature change factor and humidity change factor in proportion to the inter-cell capillary spreading area, the ratio of undischarged blank pixels, and the shape distortion index. When the activated marking mode is a second marking mode, a step of up-correcting the basic sensitivity corresponding to the vibration factor and the repetitive impact factor in proportion to the frequency of occurrence of microcrack candidate patterns on the etching interface, the contrast deviation corresponding to the etching depth, and the laser etching integrity index; when the activated marking mode is a third marking mode, a step of up-correcting the basic sensitivity corresponding to the friction factor and the repetitive contact factor in proportion to the end lifting index, the label adhesion misalignment angle relative to the product reference axis, the bubble generation area within the adhesive surface, and the physical adhesion integrity index; and a step of obtaining environmental stress metadata from a logistics database linked with the identification information of the target product, including the estimated logistics transportation time after shipment of the target product, the dwell time per transportation segment, the estimated accumulated vibration per transportation method, the number of loading and unloading operations, the estimated temperature exposure profile and the estimated humidity exposure profile along the route.A step of generating a time-series stress sequence including friction stress, vibration stress, temperature change stress, humidity change stress, repetitive impact stress, and repetitive contact stress for each of multiple time intervals of the logistics post-processing by dividing the above environmental stress metadata according to chronological order; a step of inferring a degradation prediction curve including a trend of reading rate decline, a trend of contrast reduction, a trend of shape distortion increase, and a point of transition to unreadability according to the elapsed time of the logistics post-processing by inputting the basic sensitivity for each environmental factor and the time-series stress sequence as multiple variables into a machine learning-based time-series degradation prediction model pre-trained using historical data of identification codes judged unreadable in the past logistics environment, a history of changes in the past valid reading rate, a history of the past defect feature vector, and past environmental stress metadata; and a step of calculating, from the degradation prediction curve, the remaining valid time until the reading rate of the identification code drops below a preset limit reading value and the remaining quality maintenance time until the contrast or shape distortion exceeds a preset quality limit. A step of calculating a readability margin value based on the difference between the estimated logistics transport time and the remaining valid time, and calculating a quality maintenance margin value based on the difference between the estimated logistics transport time and the remaining quality maintenance time, and then calculating a readability margin ratio and a quality maintenance margin ratio corresponding to the readability margin value and the quality maintenance margin value, respectively; a step of setting the smaller value among the readability margin ratio and the quality maintenance margin ratio as the representative margin ratio, and inputting the representative margin ratio into a surcharge function defined such that the weight increases non-linearly as the representative margin ratio decreases below a preset safety guarantee threshold to calculate a non-linear surcharge coefficient;and may include the step of finally calculating a degradation prediction score that quantitatively indicates the possibility that the identification code will transition to an unreadable state after the logistics post-processing by combining the slope of the reading rate decline derived from the degradation prediction curve, the remaining effective time, the representative margin ratio, the basic sensitivity by environmental factor, and the non-linear markup coefficient.

[0110] The step of calculating the above degradation prediction score is not merely for using the defect characteristics and structural damage status obtained at the current inspection point for the current pass / fail determination, but is a step for predicting how the readability of the identification code will deteriorate over time under environments of friction, vibration, temperature change, humidity change, repetitive impact, and repetitive contact expected in the logistics post-shipment process.

[0111] Here, the degradation prediction score may be a predictive indicator that quantitatively represents the likelihood that a specific identification code will transition to an unreadable state during subsequent transportation and handling processes, even if it is currently in a readable state. Therefore, the above step may be a predictive maintenance type quality prediction step that is distinct from a simple defect detection step or a quality judgment step at the current point in time.

[0112] The step of extracting a multidimensional defect distribution pattern indicating the damage location, damage size, damage density, and damage propagation direction of the identification code based on the defect feature vector, the defect cluster diagram included in the zone-specific defect distribution matrix, the defect occupancy ratio and the continuous intrusion length of the defect, the valid read rate and the deduction weight penalty, is a step of integrating different types of data derived from the preceding inspection execution step and the deduction weight penalty calculation step to model the damage pattern of the barcode or 2D code in three dimensions.

[0113] Here, the multidimensional defect distribution pattern may be a complex data structure that combines simple scalar values ​​or individual indicators at the level of a one-dimensional array to comprehensively represent how severely a specific part of the identification code is damaged, whether defects are dispersed or densely clustered in a particular area, and in which direction the damage is likely to spread further when external stress is applied in the future. For example, if a long continuous intrusion length of defects is formed from the upper right to the lower left of the data encoding area and a high degree of defect clustering is calculated for the same area, the multidimensional defect distribution pattern can be extracted as a pattern that reflects this directionality and implies the possibility of cell collapse or contrast degradation spreading in a specific direction in the future. Therefore, the device can acquire pattern data for analysis that goes beyond simple defect vectors and includes the location, scale, distribution, and propagation characteristics of structural damage.

[0114] The step of projecting the multidimensional defect distribution pattern onto the previously established environmental stress transfer matrix to calculate the basic sensitivity for each environmental factor—for the friction factor, vibration factor, temperature change factor, humidity change factor, repetitive impact factor, and repetitive contact factor applied to the identification code—is a step of quantitatively quantifying how vulnerable the currently identified damaged structure is to external environmental factors.

[0115] Here, the environmental stress transfer matrix refers to correlation information pre-established in the form of a matrix or parameter map through historical data, experimental results, simulation results, or rule-based models, indicating which physical external environment causes a specific type of defect pattern to deteriorate most rapidly. Additionally, the basic sensitivity by environmental factor may be a numerical value ranging from 0 to 1 or equivalent continuous values ​​representing how easily the corresponding identification code can be damaged by each of the external stress factors—such as friction, vibration, temperature, humidity, repetitive impact, and repetitive contact—based on the extracted multidimensional defect distribution pattern. For example, if defects are concentrated near the outer edges of the code and the quiet zone on the defect distribution matrix, the probability of further damage to the code's outer structure due to contact or friction during logistics transport is high; therefore, the basic sensitivity for the friction and repetitive contact factors may be calculated as relatively high values. Accordingly, the device can analyze, by distinguishing by environmental factor, which stress will be more critical in the subsequent logistics environment, even for the same quality defect.

[0116] When the above-mentioned activated marking mode is the first marking mode, the step of up-correcting the basic sensitivity corresponding to the temperature change factor and the humidity change factor in proportion to the inter-cell capillary spreading area, the ratio of unejected blank pixels, and the shape distortion index is a step of additionally reflecting the degradation mechanism inherent to the ink-based printing method in the sensitivity calculation.

[0117] Here, a large inter-cell capillary spreading area may indicate that the ink has already diffused into surrounding cells along the paper fiber tissue or surface microstructure; thus, exposure to high temperature or high humidity environments may increase the likelihood of additional ink diffusion or boundary collapse. Additionally, if the ratio of unejected blank pixels is high, it implies that parts of the cell structure are weakly formed even at the present time, making cell boundaries more susceptible to collapse due to changes in temperature or humidity. A high shape distortion index may also imply low boundary stability of the ink pattern; therefore, the device can reflect the environmental vulnerability unique to the first marking mode by upgrading the basic sensitivity to temperature and humidity change factors using multiplicative, weighted additive, or non-linear correction methods.

[0118] When the above-mentioned activated marking mode is a second marking mode, the step of up-correcting the basic sensitivity corresponding to the vibration factor and the repetitive impact factor in proportion to the frequency of occurrence of microcrack candidate patterns at the etching interface, the contrast deviation corresponding to the etching depth, and the laser etching integrity index is a step of weighting the possibility of structural failure, which mainly occurs in the laser etching method, to the sensitivity score.

[0119] Here, a high frequency of microcrack candidate patterns at the etching interface may indicate that the heat-affected zone or etching boundary is already close to a state of microscopic damage, and a large contrast deviation corresponding to the etching depth may imply that the uniformity of the etching depth is low, suggesting that specific cells or line segments may be structurally weak. Furthermore, if the laser etching integrity index is low, repetitive mechanical shock or continuous vibration may accelerate the degradation of contrast, boundary damage, or changes in reflection characteristics of the etched area. Accordingly, the device can increase the basic sensitivity to vibration factors and repetitive shock factors to more strongly reflect physical shock vulnerability in the second marking mode in subsequent degradation prediction.

[0120] When the above-mentioned activated marking mode is the third marking mode, the step of up-correcting the basic sensitivity corresponding to the friction factor and the repetitive contact factor in proportion to the above-mentioned end lifting index, the label adhesion misalignment angle relative to the product reference axis, the bubble generation area within the adhesive surface, and the physical adhesion integrity index is a step that reflects the risk of loss of adhesive strength and detachment that mainly occurs in the label attachment method.

[0121] Here, a high edge lifting index may indicate that the label edge is already separated from the surface and can peel off more easily due to external friction, and a large label adhesion misalignment angle relative to the product reference axis may mean that the entire label is misaligned from the preset attachment position, suggesting a high likelihood of abnormal stress acting on the outer boundary. Furthermore, if the area of ​​air bubbles within the adhesive surface is large or the physical adhesion integrity index is low, label wrinkling, localized peeling, or outline deformation may progress rapidly during repeated contact or handling. Therefore, the device can clearly reflect the adhesion degradation mechanism of the third marking mode by adjusting the sensitivity to friction factors and repeated contact factors upward to the most vulnerable level.

[0122] The step of obtaining environmental stress metadata, including the estimated logistics transport time after shipment of the target product, the dwell time per transport segment, the estimated cumulative vibration per transport method, the number of loading and unloading operations, the estimated temperature exposure profile and the estimated humidity exposure profile along the route, from a logistics database linked with the identification information of the target product, is a step of collecting future external environmental conditions that the target product will experience from leaving the factory until reaching the final destination.

[0123] Here, the logistics database may be a DB or linked system integrated with an ERP, logistics tracking system, transport schedule management server, or external delivery platform via API. Additionally, the environmental stress metadata may be structured data including transport planning information, storage segment information, loading and unloading history, climate information by route, vibration characteristic information by transport method, and customer delivery condition information. For example, schedule information such as a specific product passing near the equator via a sea transport vessel for 31 days, undergoing a total of 5 unloading operations and 3 days of unpaved road truck transport, can be obtained in the form of a metadata structure. Accordingly, the device can reflect individualized environmental conditions corresponding to the actual product logistics route in subsequent predictions, rather than using simple average environmental assumptions.

[0124] The step of dividing the above environmental stress metadata according to chronological order to generate a time-series stress sequence including friction stress, vibration stress, temperature change stress, humidity change stress, repetitive impact stress, and repetitive contact stress for each of multiple time intervals of the logistics post-processing is a step of converting the acquired environmental schedule information into time-axis-based prediction input data.

[0125] Here, a time-series stress sequence refers to sequential data in which the entire logistics route is subdivided into hourly, daily, or event units, and the expected types and intensities of stress for each time interval are arranged in the form of vectors or tensors. For example, the amount of repetitive impact stress and repetitive contact stress may be set high during the loading section immediately after shipment; the amount of vibration stress may be set high during the long-distance vehicle transport section; and the amount of temperature change stress and humidity change stress may be set high during the section passing through high-temperature and high-humidity areas. Therefore, the device can interpret the logistics environment not as a single average value, but as a complex stress pattern that changes sequentially over time.

[0126] The step of inferring a degradation prediction curve including the trend of reading rate decline, contrast reduction, shape distortion increase, and the point of transition to unreadability according to the elapsed time of the logistics post-processing, by inputting the basic sensitivity for each environmental factor and the time-series stress sequence as multiple variables into a machine learning-based time-series degradation prediction model pre-trained using historical data of identification codes judged unreadable in the past logistics environment, historical changes in the past valid reading rate, historical defect feature vector history, and historical environmental stress metadata, is a step of predicting future quality changes by combining the current state of the product with past similar degradation history.

[0127] Here, the time-series degradation prediction model may be a machine learning-based prediction model including RNN, LSTM, GRU, Transformer, or time-series regression models, and the model may implicitly reflect how a combination of specific defect patterns and environmental stresses forms a quality degradation curve over time by learning cases of transition to an unreadable state in actual logistics post-processing in the past. Additionally, the degradation prediction curve may be a continuous or discrete prediction function that indicates how quality indicators such as read rate, contrast, and shape distortion change along the time axis. Therefore, the device can predict not only the quality at the current point in time but also how the identification code quality will change at future points in time of 12, 24, 48 hours, or more.

[0128] The step of calculating, respectively from the above degradation prediction curve, the remaining effective time until the reading rate of the identification code drops below a preset limit readable value and the remaining quality maintenance time until the contrast or shape distortion exceeds a preset quality limit, is a step of deriving a business-meaningful survival period value from the above prediction curve.

[0129] Here, the remaining effective time may refer to the time remaining until the predicted read rate falls below the minimum allowable read threshold, and the remaining quality retention time may refer to the time remaining until at least one of contrast or shape distortion exceeds the quality tolerance range. For example, if the read rate is predicted to drop below the limit readability value at 60 hours after shipment on the degradation prediction curve, the remaining effective time may be calculated as 60 hours. Additionally, if the shape distortion is predicted to exceed the quality limit at 48 hours after shipment, the remaining quality retention time may be calculated as 48 hours. Accordingly, the device can quantitatively determine the time margin during which the corresponding identification code can withstand the entire actual logistics process, going beyond simply determining whether it is a good product.

[0130] The step of calculating a readability margin value based on the difference between the estimated logistics transport time and the remaining valid time, and calculating a quality maintenance margin value based on the difference between the estimated logistics transport time and the remaining quality maintenance time, and then calculating the readability margin ratio and the quality maintenance margin ratio corresponding to the readability margin value and the quality maintenance margin value, respectively, is a step of quantifying the margin or shortage between the predicted quality maintenance time and the actual required logistics duration.

[0131] Here, the readability margin value may be the value obtained by subtracting the estimated logistics transport time from the remaining valid time, and the quality retention margin value may be the value obtained by subtracting the estimated logistics transport time from the remaining quality retention time. For example, if the estimated logistics transport time is 72 hours and the remaining valid time is predicted to be 80 hours, the readability margin value may be 8 hours. Conversely, if the remaining quality retention time is only 60 hours, the quality retention margin value may be -12 hours. In addition, since the margin ratio may be a value normalized by dividing the above margin value by the estimated logistics transport time or a preset standard time, the device can compare the possibility of quality retention by reflecting not only the absolute time difference but also the relative margin level.

[0132] The step of setting the smaller value between the above-mentioned readability margin ratio and quality maintenance margin ratio as the representative margin ratio, and calculating a non-linear premium coefficient by inputting the representative margin ratio into a premium function defined such that the weight increases non-linearly as the representative margin ratio decreases below a preset safety guarantee threshold, is a step of calculating a risk amplification factor based on the most vulnerable condition among the calculated margins.

[0133] Here, the representative margin ratio may be a value used to represent a vulnerable indicator that is most likely to reach its limit first during actual transportation, by selecting the more unfavorable value between readability and quality maintenance. Additionally, the safety guarantee threshold may be the minimum margin level that must be secured to maintain quality stably throughout the entire logistics route. For example, a representative margin ratio of 0.3 or higher can be interpreted as having sufficient margin, but if it drops below 0.1, it may switch to an unreadable state even with very small environmental fluctuations; therefore, a non-linear surcharge factor can be set to increase rapidly through a non-linear surcharge function, such as an Exponential function. Thus, the device can reflect actual logistics quality risks where the risk level increases sharply near the threshold, rather than simple linear penalties.

[0134] The step of finally calculating a degradation prediction score, which quantitatively indicates the probability that the identification code will transition to an unreadable state after the logistics post-processing by combining the slope of decline in the read rate derived from the above degradation prediction curve, the remaining effective time, the representative margin ratio, the basic sensitivity by environmental factor, and the non-linear markup coefficient, is a final prediction step that integrates the current defect status, environmental vulnerability, the degradation rate over time, and the margin level relative to the logistics duration.

[0135] Here, the slope of the reading rate decline refers to the speed at which the reading rate decreases over time on the degradation prediction curve; the lower the representative margin ratio and the higher the non-linear markup factor, the higher the final degradation prediction score may be calculated. Additionally, if the sensitivity to specific factors among the basic sensitivities for environmental factors—for example, the friction factor and the repetitive contact factor in the case of the third marking mode—is high, the degradation prediction score may be higher even with the same logistics transport time. Therefore, the above degradation prediction score is not merely a current defect score, but rather a prediction result that comprehensively quantifies the probability that the identification code will transition to an unreadable state after the subsequent logistics process. This degradation prediction score is transmitted to the integrated management stage and can be utilized as a key control criterion to adjust the settings of currently operating marking equipment, determine whether rework is necessary, or decide whether to change the logistics route or reinforce packaging, based on the probability of future defects that are difficult for operators to perceive.

[0136] An apparatus according to one embodiment includes a processor and memory. The processor may include at least one apparatus described above through the drawings or perform at least one method described above through the drawings. The memory may store information related to the method described above or store a program in which the method described above is implemented. The memory may be volatile memory or non-volatile memory.

[0137] The processor can execute a program and control the device. The code of the program executed by the processor can be stored in memory. The device can be connected to an external device (e.g., a personal computer or a network) through an input / output device (not shown in the drawing) and exchange data.

[0138] The embodiments described above may be implemented as hardware components, software components, and / or combinations of hardware and software components. For example, the devices, methods, and components described in the embodiments may be implemented using one or more general-purpose or special-purpose computers, such as, for example, a processor, a controller, an arithmetic logic unit (ALU), a digital signal processor, a microcomputer, a field programmable gate array (FPGA), a programmable logic unit (PLU), a microprocessor, or any other device capable of executing and responding to instructions. The processing unit may execute an operating system (OS) and one or more software applications executed on said operating system. Additionally, the processing unit may access, store, manipulate, process, and generate data in response to the execution of the software. For ease of understanding, the processing unit may be described as being used as a single unit, but those skilled in the art will understand that the processing unit may include multiple processing elements and / or multiple types of processing elements. For example, the processing unit may include multiple processors or one processor and one controller. Additionally, other processing configurations, such as parallel processors, are also possible.

[0139] The method according to the embodiment may be implemented in the form of program instructions that can be executed through various computer means and recorded on a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, etc., either alone or in combination. The program instructions recorded on the medium may be those specifically designed and configured for the embodiment, or they may be those known and available to those skilled in the art of computer software. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes; optical recording media such as CD-ROMs and DVDs; magneto-optical media such as floptical disks; and hardware devices specifically configured to store and execute program instructions, such as ROM, RAM, and flash memory. Examples of program instructions include machine code, such as that generated by a compiler, as well as high-level language code that can be executed by a computer using an interpreter, etc. The hardware devices described above may be configured to operate as one or more software modules to perform the operation of the embodiment, and vice versa.

[0140] Software may include computer programs, code, instructions, or a combination of one or more of these, and may configure a processing unit to operate as desired or instruct the processing unit independently or collectively. Software and / or data may be permanently or temporarily embodied in any type of machine, component, physical device, virtual equipment, computer storage medium, or device so as to be interpreted by the processing unit or to provide instructions or data to the processing unit. Software may be distributed over networked computer systems and stored or executed in a distributed manner. Software and data may be stored on one or more computer-readable recording media.

[0141] Although the embodiments have been described above with reference to the limited drawings, those skilled in the art can apply various technical modifications and variations based on the above. For example, suitable results may be achieved even if the described techniques are performed in a different order than described, and / or if the components of the described system, structure, device, circuit, etc. are combined or assembled in a form different from described, or replaced or substituted by other components or equivalents.

[0142] Therefore, other implementations, other embodiments, and equivalents to the claims also fall within the scope of the claims set forth below. Explanation of the symbols

[0143] S100: Marking method determination step S110: Step to generate a multidimensional feature vector S120: A step of obtaining a weight matrix by marking mode, a set of allowable thresholds, and a set of exclusion rules. S130: Step for calculating basic fitness values ​​for each mode S140: Step for calculating the final process conformity score S150: Step of excluding the first marking mode from the decision candidates S160: Step of excluding the second marking mode from the decision candidates S170: Mode Selection Step S200: Marking execution step S300: Inspection execution step S400: Integrated Management Phase

Claims

Claim 1 A method for providing an automated marking and inspection solution for a surface printing process according to product material, wherein the device includes a processor, memory, a communication module, and a non-transient storage medium, and the program stored in the non-transient storage medium is executed by the processor, comprising: a marking method determination step of acquiring surface material data of a target product entering a production line and determining one of ink marking, laser marking, and label attachment optimized for the surface material data by referring to a pre-established marking mapping table by material; a marking execution step of printing or attaching an identification code on the surface of the target product by activating one selected mode among a first marking mode that controls the ejection of an inkjet head according to the marking method, a second marking mode that etches the surface by controlling the output and irradiation path of a laser oscillator, and a third marking mode that attaches a physical label by driving a label dispenser; and an inspection image acquired by capturing a region of interest (ROI) on which the identification code is indicated using a vision camera, and at least one of contrast, shape distortion, and read rate of the identification code within the inspection image based on a vision reading algorithm. The method comprises: an inspection execution step for generating quality reading data including; and an integrated management step for mutually mapping control parameters applied in the marking execution step with the quality reading data and storing them in a history database, and performing automatic feedback control to change the determination logic of the marking method or control parameters corresponding to the surface material data of the target product when the quality reading data is below a preset allowable standard value; wherein the marking method determination step comprises: a step of calculating a surface roughness index, a light reflectance index, a glossiness index, and a surface energy estimate of the target product based on sensing information measured from a sensor unit placed on the production line, and generating a multidimensional characteristic vector including the surface roughness index, the light reflectance index, the glossiness index, and the surface energy estimate.A step of obtaining a marking mode-specific weight matrix, an allowable threshold set, and an exclusion rule set corresponding to each of the first to third marking modes from the material-specific marking mapping table by referring to the material code, item code, required reading grade, and production speed conditions included in the product identification information of the target product; a step of calculating a mode-specific basic suitability value including ink wettability suitability for the first marking mode of the target product, laser response suitability for the second marking mode, and label adhesion suitability for the third marking mode by performing matrix operations on the multidimensional characteristic vector and the marking mode-specific weight matrix; and a step of calculating a final process suitability score for each of the first to third marking modes by reflecting risk penalty factors, including ink smudging risk, non-adhesion risk, laser diffuse reflection risk, thermal damage risk, label peeling risk, and attachment position error risk, respectively, to the mode-specific basic suitability value. A step of excluding the first marking mode from the determination candidates to preemptively prevent ink drop, smudging, or non-adhesion defects when the surface energy estimate is below a preset threshold dyne value for securing ink wettability, or when the surface roughness index exceeds a preset roughness range that hinders uniform ink application; a step of excluding the second marking mode from the determination candidates to prevent engraving omission, contrast degradation, or damage to surrounding optical equipment due to diffuse reflection of the laser beam when the light reflectance index or gloss index is above a preset high-reflectance threshold, or when the thermal sensitivity coefficient corresponding to the material code is above a preset thermal damage criterion;The method comprises a mode selection step in which the marking mode with the highest final process suitability score is selected as the priority candidate method, wherein if the score difference between the priority candidate method and the next-ranked method is less than a preset competition threshold, a correction score reflecting past reading success rates, defect rates, and rework rates for the same material group or similar surface characteristic group is additionally calculated by referring to a history database, and the marking mode with the higher correction score is determined as the final marking method, wherein if both the first marking mode and the second marking mode are excluded from the determination candidates, the third marking mode is determined as the final marking method; and the inspection execution step comprises: a step of controlling the irradiation angle, irradiation direction, visible light wavelength band, polarization state, and light intensity of a multi-wavelength illumination module based on the activated marking mode, expected marking location information of the target product, and optical response characteristics of the target product surface, and acquiring a diffuse reflection inspection image in which specular reflection components are suppressed; A step of generating an aligned ROI image by performing lens distortion correction, perspective correction, luminance flattening, highlight removal, and image alignment on the diffuse reflection inspection image, and extracting a region of interest marked with an identification code by referencing at least one of the outer contour, reference mark, reference hole, corner, and product reference axis of the target product; a step of segmenting the cell area, code boundary area, and background area of ​​the identification code by calling an image processing recipe pre-mapped to the activated marking mode and selectively applying at least one morphological filter among an adaptive binarization algorithm, a contrast enhancement filter, an edge enhancement filter, and dilation, erosion, opening, and closing to the aligned ROI image;When the activated marking mode is a first marking mode, the step of measuring the distribution density of satellite droplets, the capillary spreading area between cells, the ratio of unejected blank pixels, the deviation in ink diffusion directionality, and the boundary roughness index of the code outer edge based on the cell area and the code boundary area, and deriving a first mode index including an ink spreading defect index and a shape distortion index using the distribution density of satellite droplets, the capillary spreading area between cells, the ratio of unejected blank pixels, the deviation in ink diffusion directionality, and the boundary roughness index of the code outer edge; When the above-mentioned activated marking mode is a second marking mode, the method comprises the step of measuring the discoloration spectrum deviation of the heat-affected zone, the frequency of occurrence of microcrack candidate patterns at the etching interface, the optical response non-uniformity at the laser beam overlap point, and the contrast deviation corresponding to the carbonized edge width and etching depth using the optical response difference due to multi-angle illumination and the change in reflected luminance by wavelength band, and deriving a second mode index including a laser etching integrity index and a contrast index using the discoloration spectrum deviation of the heat-affected zone, the frequency of occurrence of microcrack candidate patterns at the etching interface, the optical response non-uniformity at the laser beam overlap point, and the contrast deviation corresponding to the carbonized edge width and etching depth; If the activated marking mode is a third marking mode, the method comprises the step of performing contour tracking and edge profile analysis together with code outline information required for reading the identification code to measure an end lifting index based on changes in the radius of curvature of the label edge, a label adhesion misalignment angle relative to the product reference axis, an area of ​​bubble occurrence within the adhesive surface, a wrinkle distribution density, and a positional deviation of the label outline, and deriving a third mode index including a physical adhesion integrity index using the end lifting index, the label adhesion misalignment angle relative to the product reference axis, the area of ​​bubble occurrence within the adhesive surface, the wrinkle distribution density, and the positional deviation of the label outline;A step of calculating a basic read rate value of the cell area of ​​the identification code based on international standard barcode specifications, generating a defect feature vector by normalizing the detailed defect measurement value derived corresponding to the activated marking mode and the mode index corresponding to the activated marking mode, and calculating a point deduction weight penalty based on the severity of each item included in the defect feature vector; a step of calculating an effective read rate by reflecting the point deduction weight penalty to the basic read rate value; a step of analyzing a multidimensional defect distribution pattern from the defect feature vector to calculate a degradation prediction score that quantifies physical vulnerability to friction, vibration, and temperature and humidity change environments in the logistics post-process; a step of calculating an initial read failure risk based on the difference between the effective read rate and a preset target read threshold value, and calculating a risk of conversion to unreadable status by reflecting the degradation prediction score to the initial read failure risk. and a step of generating final quality read data by mapping the valid read rate, defect feature vector, degradation prediction score, and unreadable conversion risk; wherein the step of calculating the deduction weight penalty comprises: aligning a symbolage layout corresponding to the specification information of the identification code on a reference coordinate system, and then spatially dividing the symbolage layout to set a plurality of topological importance zones including a finder pattern area, a timing pattern area, a data encoding area, and a quiet zone area, wherein if at least one of an alignment pattern area, a version information area, and a format information area exists according to the identification code specification information, it is set as an additional auxiliary importance zone;Based on the spatial data of the aligned ROI image and the segmented cell area and code boundary area, the pixel coordinates of detailed defect measurements, defect area, defect outer shape, and defect cluster map are mapped to the reference coordinate system to generate a zone-specific defect distribution matrix including the defect occupancy area, defect occupancy ratio, defect cluster map, defect center coordinate distribution, and continuous intrusion length of defects for each of the topological importance zones; identifying an error correction level corresponding to the international standard barcode specification of the identification code, and calculating an error correction exhaustion rate representing the ratio of codewords damaged by defects to the total error correction restoration capacity of the identification code based on the defect occupancy ratio and the continuous intrusion length of defects within the data encoding area; Regarding the defect distribution matrix by zone, a first weight of the highest grade is assigned to defects occurring in the finder pattern area and timing pattern area, reflecting the risk of direct connection to unrecognizability due to loss of decoding reference axis or code alignment failure; a second weight is assigned to defects occurring in the quiet zone area, reflecting the level of background separation margin infringement; and an auxiliary weight is assigned to defects occurring in the alignment pattern area, version information area, or format information area, reflecting the level of code consistency degradation; and regarding defects occurring in the data encoding area, a third weight is calculated by inputting the error correction exhaustion rate, defect clustering, and the continuous intrusion length of the defect into a weighting function defined such that the weight increases non-linearly based on a threshold point as the error correction exhaustion rate approaches a preset limit restoration rate.A step of calculating a penalty component value for each zone by combining the corresponding weight, defect occupancy area, and defect occupancy ratio for each of the finder pattern area, timing pattern area, data encoding area, quiet zone area, and auxiliary importance zone, and calculating a first penalty value by weighted summing the penalty component values ​​for each zone; a step of calculating a synergy coefficient that reflects the reading interference effect due to mutual interference between multiple defects when, within the defect feature vector, at least two detailed defect measurements among the unejected blank pixel ratio, inter-cell capillary bleeding area, contrast deviation corresponding to etching depth, frequency of occurrence of microcrack candidate patterns at the etching boundary, end lifting index, and label adhesion misalignment angle relative to the product reference axis overlap across adjacent topological importance zones, or when the clustering of detailed defect measurements that directly impair the contrast of the code exceeds a preset threshold. A method for providing an automated marking and inspection solution for a surface printing process according to a product material, comprising the step of deriving a final weighted penalty by reflecting a synergy coefficient in the first penalty value, wherein an additional upper limit coefficient is applied so that the penalty value before and after the application of the synergy coefficient does not exceed a preset stability limit; Claim 2 delete Claim 3 delete

Images