A label pattern local pre-distortion generation method for corner areas

Abstract

This application relates to the field of image processing technology, and in particular to a method for generating local pre-distortion of label patterns for corner areas. The method includes acquiring the three-dimensional contour surface of the label-bearing object and label design data; constructing a width sampling line with normal, tangential, and arc length positions; calculating the tangential angle change, normal change, and arc length deviation of adjacent points based on the surface points, generating a curvature abrupt change intensity sequence; separating and obtaining candidate corner areas based on the curvature abrupt change intensity sequence; calculating the deformation influence of grid points, constructing a local distortion weight field; and generating a reverse pre-distortion vector field based on the local distortion weight field. This application establishes a curvature analysis grid, mapping the label's two-dimensional design data to the three-dimensional contour surface of the bearing object, and identifies corner areas, transition areas, and planar areas based on tangential changes, normal changes, and arc length deviations, thus accurately determining the actual deformation influence range when the label is attached across corners.

Description

Technical Field

[0001] This application relates to the field of image processing technology, specifically a method for generating local pre-distortion of label patterns for corner areas. Background Technology

[0002] With the increasing application of product identification, traceability labels, anti-counterfeiting labels, and QR code labels on packaging containers, electronic device casings, curved components, and other load-bearing objects, labels often need to be applied across corner areas. On flat surfaces, the label pattern can usually maintain its original two-dimensional design. However, in corner areas, the surface curvature of the load-bearing object changes significantly. During the pressing process, the label material experiences localized stretching, compression, slippage, and springback, causing the applied pattern to shift or deform relative to its original design position. This is particularly problematic as it can easily affect the continuity and recognition stability of corner lines, QR codes, barcodes, and other pattern elements.

[0003] Existing label pattern preprocessing methods typically adjust the pattern as a whole based on the designed corner line or empirical compensation, which fails to accurately reflect the location of curvature abrupt changes in the actual three-dimensional contour of the object being supported, and also fails to reflect the influence of adhesion tension, adhesive strength, and material springback on the actual bending center. When the actual bending center does not match the designed corner line, problems such as insufficient compensation in the corner area, discontinuous compensation in the transition area, and miscompensation in the planar area easily occur, resulting in misalignment, broken lines, compression, or recognition difficulties in the label pattern after application. Therefore, a label pattern generation method is needed that can dynamically determine the influence range of the corner based on the three-dimensional contour of the object being supported and the application process, and perform local pre-distortion processing only on the corner area. Summary of the Invention

[0004] To address the aforementioned issues, this application provides a method for generating local pre-distortion of label patterns for corner areas. By establishing a curvature analysis mesh, the two-dimensional design data of the label is mapped to the three-dimensional contour surface of the supporting object. Based on tangential changes, normal changes, and arc length deviations, corner areas, transition areas, and planar areas are identified, accurately determining the actual deformation impact range when the label is applied across corners. Simultaneously, this application combines application tension, adhesive layer bonding strength, label material springback coefficient, and pressing direction to calculate dynamic migration, forming a dynamic corner polygonal line. This ensures that the pre-distortion center aligns with the predicted bending center, avoiding the compensation position offset, insufficient corner area compensation, and incorrect planar area compensation problems caused by compensating only according to a fixed design corner line in existing technologies.

[0005] This application adopts the following technical solution: a method for generating local pre-distortion of label patterns for corner areas, which obtains the three-dimensional contour surface of the label bearing object and label design data, establishes a curvature analysis grid, maps each grid point in the curvature analysis grid to the three-dimensional contour surface of the bearing object, obtains the corresponding surface points, and sorts the surface points at the same width position along the label attachment direction to construct a width sampling line with normal, tangential and arc length positions;

[0006] Based on the surface points, the changes in tangential angle, normal angle, and arc length deviation of adjacent points are calculated to generate a curvature change intensity sequence. Candidate corner regions are then separated based on the curvature change intensity sequence.

[0007] By combining the candidate corner center of the candidate corner area and the attachment process parameters, the dynamic migration amount of the predicted bending center relative to the candidate corner center is calculated, and the dynamic corner line is determined based on the dynamic migration amount.

[0008] Calculate the deformation influence of grid points, and search for transition boundaries that satisfy the preset cutoff conditions from the dynamic corner broken line along each width sampling line to both sides to obtain the negative side transition width and the positive side transition width. Generate a dynamic transition influence band based on the negative side transition width and the positive side transition width, and determine the area outside the dynamic transition influence band as the plane preservation area.

[0009] Based on the dynamic corner polyline, the dynamic transition influence zone, and the label design outline, a local distortion weight field is constructed.

[0010] Based on the local distortion weight field, an inverse predistortion vector field is generated, and based on the inverse predistortion vector field, local inverse resampling is performed on the original label pattern to output a predistorted label pattern that only affects the dynamic transition influence zone.

[0011] As a further description of the above technical solution: the method for obtaining the corresponding surface points includes:

[0012] The u-direction is established based on the label attachment direction stored in the label design data, and the v-direction is established based on the label width direction stored in the label design data, forming a two-dimensional coordinate system for the label.

[0013] Curvature analysis grids are generated inside the label design outline according to preset longitudinal sampling intervals and preset transverse sampling intervals; for each grid point, the label center attachment line stored in the label design data is obtained, and the corresponding three-dimensional path points, surface normal vectors and path tangent vectors on the label center attachment line are read to calculate the three-dimensional width direction vector.

[0014] Based on the three-dimensional path point and the three-dimensional width direction vector, the initial three-dimensional corresponding point of the grid point is determined, and the initial three-dimensional corresponding point is projected onto the three-dimensional contour surface of the bearing object to obtain the corresponding surface point.

[0015] As a further description of the above technical solution: the steps for constructing the width sampling line include:

[0016] Surface points at the same width position are arranged in order along the u direction to form the j-th width sampling line. For each surface point in the width sampling line, the surface normal vector, local tangent vector, and arc length position are calculated, thus forming a width sampling line with normal, tangent, and arc length positions.

[0017] As a further description of the above technical solution: the method for generating the curvature change intensity sequence includes:

[0018] Calculate the tangential and normal changes for each surface point;

[0019] Calculate the arc length deviation for each surface point;

[0020] The curvature abrupt change intensity is obtained by weighted summation of the tangential change, normal change, and arc length deviation.

[0021] As a further description of the above technical solution: the method for obtaining candidate corner regions based on curvature change intensity sequence separation includes:

[0022] Set a first curvature threshold and a second curvature threshold, wherein the first curvature threshold is less than the second curvature threshold;

[0023] When the curvature abrupt change intensity is less than the first curvature threshold, the corresponding grid point is marked as a plane candidate point;

[0024] When the curvature abrupt change intensity is greater than or equal to the first curvature threshold and less than or equal to the second curvature threshold, the corresponding grid point is marked as a transition candidate point;

[0025] When the curvature change intensity is greater than the second curvature threshold, the corresponding grid point is marked as a corner candidate point, and consecutive adjacent corner candidate points are connected and merged to obtain a candidate corner region.

[0026] As a further description of the above technical solution: the method for calculating the dynamic migration amount includes:

[0027] Read the curvature change intensity in the candidate corner region on each width sampling line, and determine the u-direction coordinate corresponding to the sampling point with the largest curvature change intensity as the candidate corner center;

[0028] Obtain the bonding process parameters, including bonding tension, adhesive layer bonding strength and label material springback coefficient. Normalize the bonding process parameters to obtain the bending center migration coefficient.

[0029] Calculate the width of the candidate corner region in the u-direction on the corresponding width sampling line;

[0030] The dynamic migration amount is calculated based on the width in the U direction, the migration coefficient of the bending center, and the motion direction of the pressing mechanism stored in the label design data.

[0031] As a further description of the above technical solution: the method for determining the dynamic corner polygonal line includes:

[0032] The dynamic corner center is calculated based on the candidate corner center and the dynamic migration amount. The coordinates of the dynamic corner center on adjacent width sampling lines are smoothed. The smoothed dynamic corner center coordinates are connected along the v direction to form a dynamic corner polyline.

[0033] As a further description of the above technical solution: the step of determining the dynamic transition influence band includes:

[0034] For any grid point inside the label design outline, calculate the signed distance of that grid point relative to the u-direction of the dynamic corner polyline;

[0035] Based on the arc length position corresponding to the grid point and the three-dimensional surface arc length increment between adjacent surface points, the three-dimensional surface arc length increment is compared with the design sampling distance in the label's two-dimensional coordinate system to calculate the local strain estimate.

[0036] The deformation effect is calculated based on the curvature abrupt change intensity and the local strain estimate.

[0037] Along each width sampling line, starting from the center of the smooth dynamic corner, search towards the negative side and the positive side of the u direction respectively. When the deformation influence of a consecutive preset number of sampling points is less than the preset transition cutoff threshold, determine the transition boundary point on the corresponding side, and form a dynamic transition influence zone based on the transition boundary points on both sides.

[0038] As a further description of the above technical solution: the step of constructing the local distortion weight field includes:

[0039] Within the valid pattern area defined by the label design outline, read the pattern elements and mark QR codes, barcodes, fine line textures, and continuous lines across corners as pattern-sensitive elements;

[0040] For each pattern-sensitive element, its element boundary, element type, and minimum feature width are recorded, and the element boundary is expanded outward based on the minimum feature width to form a sensitive influence area;

[0041] The basic distortion weight of each grid point is calculated based on dynamic corner polylines and dynamic transition influence zones;

[0042] The pattern sensitivity coefficient of each grid point is determined based on the aforementioned sensitive influence area;

[0043] By fusing the basic distortion weights and the pattern sensitivity coefficients, the local distortion weights of each grid point are obtained, and the local distortion weight field is formed by the local distortion weights of all grid points inside the label design outline.

[0044] As a further description of the above technical solution: the step of generating the inverse predistortion vector field includes:

[0045] Read the local distortion weight, local strain estimate, dynamic migration, smoothed dynamic corner center coordinates, and the signed distance of the grid point relative to the u-direction of the dynamic corner polyline for each grid point;

[0046] The predicted deformation in the u-direction is calculated based on the local strain estimate and the dynamic migration.

[0047] The predicted deformation in the v direction is calculated based on the change in the dynamic corner line along the label width direction and the signed distance in the u direction.

[0048] The predicted deformation in the u-direction and the predicted deformation in the v-direction constitute the attachment deformation prediction vector field. The attachment deformation prediction vector field is reverse-transformed according to the local distortion weight to obtain a discrete reverse pre-distortion vector. The discrete reverse pre-distortion vector is interpolated to generate a reverse pre-distortion vector field covering the inside of the label design outline.

[0049] The beneficial effects of this application are as follows:

[0050] This application establishes a curvature analysis mesh to correspond the two-dimensional design data of the label with the three-dimensional contour surface of the supporting object. Based on tangential changes, normal changes, and arc length deviations, it identifies corner areas, transition areas, and planar areas, accurately determining the actual deformation impact range when the label is applied across corners. Simultaneously, by combining the application tension, adhesive layer bonding strength, label material springback coefficient, and pressing direction, it calculates the dynamic migration amount, forming a dynamic corner polygonal line. This ensures that the pre-distortion center aligns with the predicted bending center, avoiding the compensation position offset, insufficient corner area compensation, and incorrect planar area compensation problems caused by compensating only according to a fixed design corner line in existing technologies.

[0051] Furthermore, a local distortion weight field is constructed based on the dynamic transition influence zone, and combined with the distribution of sensitive pattern elements such as QR codes, barcodes, fine line textures, and cross-corner continuous lines, differential pre-distortion compensation is applied to different regions and different pattern types. Thus, the inverse pre-distortion vector field only acts on the dynamic transition influence zone, so that the pattern can offset the deformation caused by local stretching, compression, slippage, and springback after actual application, improving the line continuity, boundary integrity, and QR code and barcode recognition stability of the cross-corner label pattern, while keeping the planar area pattern from being over-processed. Attached Figure Description

[0052] The present application will be further explained below with reference to the accompanying drawings and embodiments:

[0053] Figure 1 A flowchart of a method for generating local pre-distortion of label patterns for corner areas provided in this application;

[0054] Figure 2 A flowchart of the method for obtaining corresponding surface points provided in this application;

[0055] Figure 3 A flowchart of the method for generating curvature mutation intensity sequences provided in this application. Detailed Implementation

[0056] To make the technical means, inventive features, objectives, and effects of this application easier to understand, the application is further described below with reference to specific illustrations. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0057] Please see Figures 1-3 This application provides a technical solution: a method for generating local pre-distortion of label patterns for corner areas, comprising:

[0058] S1: Obtain the 3D contour surface of the label-bearing object and the label design data, establish a curvature analysis mesh, map each mesh point in the curvature analysis mesh to the 3D contour surface of the bearing object to obtain the corresponding surface points, and sort the surface points at the same width position along the label attachment direction to construct a width sampling line with normal, tangential, and arc length positions; in some implementations, the implementation steps include:

[0059] The three-dimensional contour surface can be directly read from the three-dimensional design model of the object being carried, or it can be obtained through structured light scanning, laser contour scanning, or three-dimensional visual reconstruction.

[0060] Acquire label design data, which includes the label design outline Ω and the label attachment path. The attachment path includes the label attachment start point, label attachment end point, label center attachment line, pressing mechanism movement direction, label attachment direction, and label width unfolding direction. The label center attachment line is used to determine the u-direction corresponding to the label attachment direction, and the pressing mechanism movement direction is used to determine the subsequent cornering and zigzag migration direction. The label attachment start point and label attachment end point are used to limit the effective path range of the label center attachment line and to determine the sampling start and end boundaries in the u-direction.

[0061] Methods for establishing a curvature analysis mesh within the label coordinate system and mapping each mesh point to a 3D contour surface to form surface points with normal, tangential, and arc length positions include:

[0062] Establish by label attachment direction Direction, established along the width of the label. Direction, forming a two-dimensional coordinate system for the label, obtaining the label design outline Ω, and obtaining the interior of the label design outline Ω. = (u, v);

[0063] According to the preset longitudinal sampling interval and preset horizontal sampling interval A curvature analysis mesh is generated within the label attachment area, and the mesh points are represented as follows: Where i represents the sampling sequence number in the u direction and j represents the sampling sequence number in the v direction. . ;

[0064] Each grid point G ij Projecting the label's target attachment position onto the 3D contour surface of the supporting object, we obtain the corresponding surface points. And read the surface normal vector of that surface point. ;

[0065] in, The sampling interval in the label attachment direction. The sampling interval is along the width of the label. N represents the coordinates of the i-th sampling point in the u direction and the j-th sampling point in the v direction in three-dimensional space, respectively; ij P represents ij The unit normal vector at that location.

[0066] Specifically, each grid point G ij Projecting the label's target attachment position onto the 3D contour surface of the supporting object, we obtain the corresponding surface points. The methods include:

[0067] First, read the line attached to the center of the label and u. i The corresponding 3D path point Ci Then read the surface normal vector at that path point. and path tangent vector And calculate the three-dimensional width direction vector at that location. ;

[0068] The formula for calculating the three-dimensional width direction vector is: ;

[0069] in, This represents the unit direction vector on the three-dimensional contour surface of the object being carried, corresponding to the width direction of the label. Represents path point C i The unit normal vector at that location; Represents path point C i The local tangent vector along the attachment path; × represents the vector cross product; |·| represents the vector length.

[0070] According to C i and Calculate grid points The initial corresponding point C in three-dimensional space i +v i ·V c i The initial corresponding point is then projected onto the three-dimensional contour surface S of the supporting object to obtain the surface point P. ij ; ;in, express G ij The three-dimensional surface points mapped onto the three-dimensional contour surface S of the carrying object; ProjS represents the projection operation of projecting spatial points onto the three-dimensional contour surface S of the carrying object.

[0071] For surface point P at the same width position ij Arrange them in order along the u-direction to form the j-th width sampling line, and calculate the surface normal vector N for each surface point in this width sampling line. ij Local tangent vector and arc length position This forms a width sampling line with normal, tangential, and arc length positions.

[0072] Specifically, the local tangent vector T ij The calculation method is as follows: ;in, Represents the local tangent vector of the i-th surface point along the attachment direction; Indicates the next surface point; Indicates the previous surface point; the arc length position The calculation method is as follows: ;in, This indicates the distance from the starting point of the j-th width sampling line to P. ij The cumulative arc length of the three-dimensional surface; r represents the cumulative sequence number; and This represents two adjacent surface points.

[0073] S2: Calculate the tangential change, normal angle change, and arc length deviation of adjacent points based on the surface points, generate a curvature abrupt change intensity sequence, and separate and obtain candidate corner regions based on the curvature abrupt change intensity sequence; in some embodiments, the implementation steps include:

[0074] Methods for generating curvature change intensity sequences include:

[0075] For each surface point P ij Calculate the tangential change and normal change ;

[0076] For each surface point P ij Calculate the arc length deviation;

[0077] Based on the tangential change Change in normal direction The curvature abrupt change intensity is obtained by weighted summation of the arc length deviation and the total curvature deviation.

[0078] Specifically, tangential change The calculation formula is: ;in, This indicates the change in the tangential angle between adjacent sampling positions along the attachment direction; Represents the local tangent vector of the current surface point; "·" represents the local tangent vector of the previous surface point; "·" represents the vector dot product.

[0079] normal change The calculation formula is: ;in, This represents the change in the surface normal angle between adjacent sampling locations; This represents the unit normal vector of the current surface point; This represents the unit normal vector of the previous surface point;

[0080] The formula for calculating the arc length deviation is: ;in, This indicates the degree of deviation of the arc length of adjacent three-dimensional surfaces relative to the sampling distance of the label's two-dimensional design; This represents the three-dimensional surface arc length increment between two adjacent surface points; This represents the vertical sampling interval in the label's two-dimensional coordinate system.

[0081] The formula for calculating the intensity of curvature abrupt change is: ;in, Represents surface point P ij The intensity of the curvature abrupt change at the corresponding position; Indicates the reference value for tangential variation; Indicates the reference value for normal change; This indicates that the arc length deviates from the reference value; , , Let represent the weight coefficients, and satisfy: + + =1, determined based on the contribution of tangential change, normal change and arc length deviation in the calibration sample to the differentiation of the labeled corner core area; in one implementation, the normalized difference degree of the three types of indicators between the plane sample and the corner sample is calculated respectively, and α, β and γ are determined according to the proportion of normalized difference degree. , and The results were calculated from calibration samples of similar load-bearing objects.

[0082] Methods for obtaining candidate corner regions based on curvature change intensity sequence separation include:

[0083] A first curvature threshold and a second curvature threshold are preset, and the first curvature threshold is less than the second curvature threshold;

[0084] When the curvature abrupt change intensity is less than the first curvature threshold, the corresponding grid point G will be... ij Mark as a candidate point in the plane;

[0085] When the curvature abrupt change intensity is greater than or equal to the first curvature threshold and less than or equal to the second curvature threshold, the corresponding grid point G is... ij Mark as a transition candidate point;

[0086] When the curvature abrupt change intensity is greater than the second curvature threshold, the corresponding grid point G will be... ij Marked as a corner candidate point.

[0087] Connecting consecutive adjacent corner candidate points yields candidate corner regions. Specifically, in the curvature analysis grid, if two corner candidate points are adjacent in the u or v direction and the interval between them does not exceed one sampling interval, they are considered connected; the four-neighborhood connectivity rule is preferred.

[0088] Specifically, the first curvature threshold is used to distinguish between planar areas and non-planar influence areas. It is set as follows: First, select several planar sample points on similar load-bearing objects that have been confirmed to belong to the planar attachment area. Following the same sampling interval and calculation rules as the formal calculation, obtain the curvature abrupt change intensity corresponding to each planar sample point. Since planar areas theoretically do not have obvious transitions, but actual 3D scanning errors, minor surface undulations, and sampling discrepancies can cause slight fluctuations in the curvature abrupt change intensity, the upper limit of normal fluctuations in the curvature abrupt change intensity in the planar samples is used as the first curvature threshold. Specifically, the concentration level and dispersion of the curvature abrupt change intensity corresponding to the planar samples can be statistically analyzed first. Then, a margin sufficient to cover normal fluctuations is superimposed on this concentration level, making the first curvature threshold higher than the conventional measured fluctuation value of the planar area. Therefore, when the curvature abrupt change intensity of a certain grid point is lower than the first curvature threshold, it indicates that the tangential change, normal change, and arc length deviation at that location are all within the normal fluctuation range of the plane, and it can be determined as a candidate planar point. When it reaches or exceeds the first curvature threshold, it indicates that the location has been affected by corners or transitional morphology and is no longer treated as a pure planar area.

[0089] The second curvature threshold is used to distinguish between the transition zone and the corner core zone. It is set as follows: First, select several corner sample points on similar load-bearing objects that have been confirmed to be at the corner core location. Then, obtain the curvature abrupt change intensity corresponding to each corner sample point according to the same rules as the formal calculation. Since the corner core zone typically exhibits rapid changes in the tangential direction, a significant change in the surface normal, and an increased deviation between the surface arc length and the two-dimensional design distance, its curvature abrupt change intensity should be significantly higher than that of the planar area and the ordinary transition zone. When setting the second curvature threshold, the curvature abrupt change intensity that can stably characterize the corner core deformation in the corner sample is used as the benchmark, and the second curvature threshold is ensured to be greater than the first curvature threshold. Therefore, when the curvature abrupt change intensity of a certain grid point is between the first and second curvature thresholds, it indicates that the location has been affected by the corner but has not yet entered the main bending center, and can be identified as a transition candidate point; when it reaches or exceeds the second curvature threshold, it indicates that the location has entered the corner core deformation area, and can be identified as a corner candidate point.

[0090] Optional, first curvature threshold :

[0091] ;in and These are the mean and standard deviation of the curvature abrupt change intensity of the planar sample, respectively.

[0092] Second curvature threshold :

[0093] =max( +ΔKmin, - );in and These represent the mean and standard deviation of the curvature abrupt change intensity of the core samples at the corner, respectively, while ΔKmin represents the standard deviation of the planar samples. .

[0094] In this embodiment, by establishing a curvature analysis mesh within the effective attachment range defined by the label design contour and mapping the mesh points to the three-dimensional contour surface of the supporting object, it is possible to identify planar areas, candidate transition areas, and candidate corner areas based on the tangential changes, normal changes, and arc length deviations of surface points. This eliminates the reliance on manually set fixed corner lines for corner area determination, instead ensuring a match with the actual three-dimensional shape of the supporting object. Therefore, this solves the problem of pre-distortion application area offset caused by inaccurate corner position identification in existing technologies, improving the spatial targeting of subsequent pattern compensation.

[0095] S3: Combining the candidate corner center of the candidate corner area and the attachment process parameters, calculate the dynamic migration amount of the predicted bending center relative to the candidate corner center, and determine the dynamic corner line based on the dynamic migration amount;

[0096] In some implementation methods, the implementation steps include:

[0097] In the j-th width sampling line, read the intensity K of all curvature abrupt changes located within the candidate corner region. ij The u-direction coordinates corresponding to the sampling point with the greatest curvature abrupt change intensity are determined as candidate corner centers. ;

[0098] The bonding process parameters, including bonding tension T, adhesive layer bond strength H, and label material resilience coefficient E, are obtained. These parameters are then normalized to obtain the bending center migration coefficient. ;

[0099] Specifically, the formula for calculating the bending center migration coefficient is as follows:

[0100] ;

[0101] in, This represents the bending center migration coefficient on the j-th width sampling line; Indicates the adhesion tension; Indicates the adhesive strength; Indicate the resilience coefficient of the label material; , and These represent reference values ​​for the corresponding parameters, obtained statistically from multiple sets of historical qualified samples under the same bearing object, label material, and qualified application process conditions. The average or median value of the application tension T, adhesive layer bonding strength G, and label material resilience coefficient R in the qualified samples are taken as the application tension reference value. Reference values ​​for adhesive layer bonding strength Reference value for the resilience coefficient of label materials ; , and Denotes the migration weight coefficients, and satisfies + + =1, , and The values ​​are determined by the attached calibration samples; the deviation of the attachment tension, the deviation of the adhesive layer bonding strength, and the deviation of the material rebound coefficient in the calibration samples are used as inputs, and the measured migration ratio is used as output. The values ​​are determined by least squares fitting, and the fitted coefficients are normalized to non-negative weights.

[0102] It should be noted that since the greater the adhesive strength, the less likely the label material is to slip, the initial adhesive strength is indicated by a negative sign.

[0103] The adhesion tension T is acquired in real time by a tension sensor on the label unwinding shaft, traction roller, or pre-application guide roller; the adhesive layer bonding strength G can be obtained by performing an initial adhesion test or peel test on the same batch of label materials on the surface of a similar carrier object. Specifically, after a preset pressing pressure and a preset dwell time, the peeling force required to peel a unit width label from the surface of the carrier object is measured, and this peeling force is taken as the adhesive layer bonding strength G; the label material springback coefficient R is obtained by performing a standard bending and release test on the same batch of label materials. Specifically, the label material is bent to a shape consistent with the corner radius of the carrier object and held for a preset time. After release, its recovery angle or recovery displacement is measured, and the ratio of the recovery angle to the initial bending angle, or the ratio of the recovery displacement to the initial bending displacement, is taken as the label material springback coefficient R.

[0104] Calculate the width of the candidate corner region in the u direction on the j-th width sampling line. :

[0105] The formula for calculating the directional width is as follows: ;in, This represents the width of the candidate corner region in the u direction on the j-th width sampling line; This represents the maximum u-direction coordinate within the candidate corner region Rc on the sampling line of this width; This represents the minimum u-direction coordinate within the candidate corner region Rc on the sampling line of this width;

[0106] The dynamic migration amount is calculated based on the width in the U direction, the migration coefficient of the bending center, and the motion direction of the pressing mechanism.

[0107] Specifically, calculate the dynamic migration amount: ;in, This represents the dynamic migration amount on the j-th width sampling line; Indicates the migration direction coefficient; Indicates the width of the candidate corner area; This represents the bending center migration coefficient after amplitude limiting.

[0108] The migration direction coefficient is obtained by reading the movement direction of the pressing mechanism based on the attachment path. When the pressing mechanism moves in the direction of increasing u coordinate, the migration direction coefficient q=1; when the pressing mechanism moves in the direction of decreasing u coordinate, the migration direction coefficient q=-1.

[0109] Among them, the bending center migration coefficient after amplitude limitation Calculate using the following formula:

[0110]

[0111] in, This indicates the maximum allowable migration coefficient.

[0112] It should be noted that the maximum migration coefficient was obtained through calibration tests on similar load-bearing objects and label materials from the same batch. Specifically, under preset pressing pressure, adhesion tension, adhesion speed, and adhesive layer conditions, several label samples were actually adhered. After adhesion, the predicted bending center position on each width sampling line was determined by visual inspection, 3D scanning, or pattern deformation analysis. The offset distance between the predicted bending center position and the candidate corner center position was recorded as the measured migration distance. This measured migration distance was divided by the width of the corresponding candidate corner area to obtain the measured migration ratio. Multiple sets of measured migration ratios were statistically analyzed, and the 95th percentile value of the measured migration ratio of qualified adhered samples was taken. When the number of samples was less than 30 sets, the maximum measured migration ratio was determined as the allowable maximum migration coefficient. .

[0113] The method for determining the dynamic corner polyline includes: based on candidate corner centers. and dynamic migration Calculate the dynamic corner center, smooth the coordinates of the dynamic corner center on adjacent width sampling lines, and connect the smoothed dynamic corner center coordinates along the v direction to form a dynamic corner polyline.

[0114] Specifically, the formula for calculating the dynamic corner center is: ,in, Let represent the coordinates of the dynamic corner center on the j-th width sampling line.

[0115] Smoothing is applied to the dynamic corner centers on adjacent width sampling lines: ;

[0116] in, This represents the coordinates of the center of the dynamically smoothed corner after smoothing. Indicates the coordinates of the dynamic corner center on the previous width sampling line; The coordinates of the dynamic corner center on the next width sampling line are shown.

[0117] In this embodiment, by combining the candidate corner center of the candidate corner area and the bonding process parameters, the dynamic migration of the bending center relative to the candidate corner center is calculated and predicted, and a dynamic corner line is formed accordingly. This method can reflect the force migration and material springback effect of the label during the actual pressing process, avoiding the problem of insufficient compensation or incorrect compensation direction in the corner core area caused by using only the designed corner line as the compensation center, thereby improving the accuracy of pattern position restoration after cross-corner bonding.

[0118] S4: Calculate the deformation influence of the grid points. Search along each width sampling line from the dynamic corner polygonal line to both sides for transition boundaries where the deformation influence satisfies the preset cutoff condition, obtaining the negative and positive transition widths. Generate a dynamic transition influence band based on the negative and positive transition widths, and determine the area outside the dynamic transition influence band as the plane preservation area, excluding the plane area from the local pre-distortion compensation range. In some embodiments, the implementation steps include:

[0119] For any grid point G inside the label ij Calculate the distance of the grid point relative to the coordinates of the center of the dynamic smooth rotation angle. ;

[0120] The formula for calculating distance is: ;in, Represents grid point G ij There is a signed distance in the u direction relative to the dynamic corner polyline; This represents the coordinates of grid point Gᵢⱼ in the label attachment direction; This represents the coordinates of the smooth dynamic corner center on the j-th width sampling line. When When <0, it indicates that grid point G ij Located on the negative side of the dynamic corner break line; when When >0, it indicates that grid point G ij Located on the positive side of the dynamic corner break line Ld; when When = 0, it indicates that grid point G ij Located on the dynamic corner line.

[0121] Based on the obtained arc length position Calculate grid point G ij Local strain estimate at the corresponding location ;

[0122] The formula for calculating the local strain estimate is:

[0123]

[0124] in, Represents grid point G ij The local strain estimate of the corresponding position relative to the two-dimensional design distance of the label; This represents the increment of the three-dimensional surface arc length between two adjacent surface points on the j-th width sampling line; This represents the design distance between two adjacent sampling points in the u-direction of the label's two-dimensional coordinate system. If... >0 indicates that the location has a local stretching tendency; if <0 indicates that the location has a local compression tendency; if =0 indicates that the 3D surface arc length increment at that location is consistent with the 2D label design distance. For the first sampling point of each width sampling line, the arc length increment of the next adjacent sampling point can be used for calculation, or the first sampling point can be excluded from the calculation of local strain estimation.

[0125] Based on obtaining the intensity of curvature change and local strain estimator Perform a weighted summation to calculate the grid point G. ij Deformation influence at corresponding locations .

[0126] Optionally, the deformation influence amount The calculation formula is: .

[0127] in, The deformation influence amount represents the value at mesh point G. ij The degree to which the corresponding position is affected by corner attachment deformation; Indicates the intensity of the curvature abrupt change; Indicates the second curvature threshold; This represents the local strain estimate; Indicates the strain reference value; , This indicates the influence weight coefficient, and satisfies... + =1. Wherein, The local strain estimates can be obtained statistically from similar load-bearing objects, label materials from the same batch, and qualified affixed samples. The preferred values ​​are those from the qualified samples. | the average or median value.

[0128] Along each width sampling line, starting from the coordinates of the center of the smooth dynamic corner, the transition boundary is searched towards the negative and positive sides respectively;

[0129] When searching towards the negative side, the deformation influence of each grid point is read sequentially along the decreasing direction of the u-axis coordinate. When the deformation influence of n consecutive sampling points When all values ​​are less than the transition cutoff threshold, the sampling point closest to the center coordinates of the smooth dynamic corner in the continuous sampling points is determined as the negative side transition boundary point, and the u-direction distance between the negative side transition boundary point and the center coordinates of the smooth dynamic corner is determined as the negative side transition width.

[0130] When searching in the positive direction, the deformation influence of each grid point is read sequentially along the direction of increasing u-coordinate. When the deformation influence of n consecutive sampling points When all values ​​are less than the transition cutoff threshold, the sampling point closest to the center coordinates of the smooth dynamic corner in the continuous sampling points is determined as the positive side transition boundary point, and the u-direction distance between the positive side transition boundary point and the center coordinates of the smooth dynamic corner is determined as the positive side transition width.

[0131] A dynamic transition influence zone is formed based on the negative side transition width and the positive side transition width.

[0132] Where n is the number of continuous stable sampling points, preferably 3 to 5; the transition cutoff threshold is determined by the deformation influence Z in the planar region. ij The upper limit of normal fluctuation is determined by taking the average value of the deformation influence in the planar region plus three times the standard deviation.

[0133] Specifically, the dynamic transition influence zone is characterized as follows: ;

[0134] in, Indicates the dynamic transition influence zone; Represents grid points Located in the label design outline internal; Represents grid points There is a signed distance relative to the u-direction of the dynamic angle polyline; - This represents the transition width on the negative side of the dynamic corner polyline on the j-th width sampling line; This represents the transition width on the positive side of the dynamic corner polygon on the j-th width sampling line. The region located outside the dynamic transition influence zone Bd is defined as the plane preservation region, which does not participate in the local corner pre-distortion compensation.

[0135] In this embodiment, the dynamic transition influence zone is determined by searching for curvature attenuation and local strain release states on both sides of the dynamic corner broken line. The area outside the dynamic transition influence zone is excluded from the local pre-distortion compensation range as a plane preservation area. Thus, the pre-distortion processing is only concentrated on the corner area and its actual influence range, which can reduce the secondary deformation of the pattern caused by the incorrect compensation of the plane area, and take into account both the compensation effect of the corner area and the stability of the pattern in the plane area.

[0136] S5: Based on the acquired dynamic corner polyline, dynamic transition influence zone, and label design contour Ω, construct a local distortion weight field W; in some implementations, the implementation steps include:

[0137] Within the valid pattern area defined by the label design outline Ω, pattern elements are read, and QR codes, barcodes, fine line textures, and continuous lines spanning corners are marked as pattern-sensitive elements to obtain the original label pattern. If the label design file is a vector file, the layer information, primitive type, outer boundary, and line width of each pattern element are read; if the label design file is a bitmap file, the outer boundary and line width of each pattern element are obtained through edge extraction and connected component segmentation.

[0138] For the k-th pattern-sensitive element, denoted as Ek, its element boundary Ak, element type Tk, and minimum feature width wk are recorded. The minimum feature width of a QR code is the width of a single module, the minimum feature width of a barcode is the narrowest bar width, and the minimum feature width of fine lines and continuous lines across corners is the line width. The element boundary Ak is expanded outward based on the minimum feature width wk to form a sensitive influence region Ak′. Preferably, the outward expansion distance for QR codes and barcodes is 2wk to 4wk, and the outward expansion distance for fine lines and continuous lines across corners is 3wk to 5wk.

[0139] Based on the dynamic corner polyline and the dynamic transition influence zone Bd, calculate G for each grid point. ij Basic distortion weights ;

[0140] Specifically, the dynamic corner core half-width h on the j-th width sampling line is determined based on the candidate corner regions. i The preferred option is to satisfy K. ij h is defined as half the width of the region that is ≥K2 and continuously located near the center of a smooth dynamic corner. i ,

[0141] When grid point G ij When located within the dynamic corner core area: ; ;

[0142] When grid point G ij When located within the negative transition zone of the dynamic corner break line: ;

[0143] When grid point G ij When located within the positive transition zone of the dynamic corner line: ;

[0144] When grid point G ij When located outside the dynamic transition influence zone: =0

[0145] in, Indicates the basic distortion weight; Represents grid point G ij There is a signed distance in the u direction relative to the dynamic corner polyline; Indicates the dynamic corner core half-width; - This represents the transition width on the negative side of the dynamic corner polyline on the j-th width sampling line; This represents the transition width on the positive side of the dynamic corner polygon on the j-th width sampling line. Through the aforementioned linear attenuation method, the basic distortion weight can be gradually reduced from the dynamic corner core region to the plane-maintaining region, with low computational cost and ease of implementation.

[0146] Based on the sensitive influence region Ak′ of the pattern-sensitive elements, determine the G of each grid point. ij Pattern sensitivity coefficient Se ij ;

[0147] Specifically, if grid point G ij If it is not located within the sensitive influence area of ​​any pattern-sensitive element, then... =0; if grid point G ij If the element is located within the sensitive influence area of ​​one or more pattern-sensitive elements, then the sensitivity coefficient corresponding to the highest sensitivity level is taken as the sensitivity coefficient. ;

[0148] In this system, QR codes and barcodes correspond to the first sensitivity coefficient, fine line textures correspond to the second sensitivity coefficient, continuous lines across angles correspond to the third sensitivity coefficient, and ordinary pattern elements correspond to zero sensitivity coefficient. The first sensitivity coefficient is greater than the second sensitivity coefficient, and the second sensitivity coefficient is greater than the third sensitivity coefficient. Optionally, the first sensitivity coefficient can be set to 1, the second sensitivity coefficient to 0.7, the third sensitivity coefficient to 0.5, and ordinary pattern elements to 0.

[0149] Ordinary pattern elements refer to filled patterns, color blocks, or decorative patterns that are not required to be continuously recognized across corners, excluding QR codes, barcodes, fine line textures, and continuous lines across corners.

[0150] Integrating basic distortion weights and pattern sensitivity coefficient iThe local distortion weights are obtained. The local distortion weight field W is composed of the local distortion weights corresponding to all grid points inside the label design outline.

[0151] The fusion method is as follows: ;in, Indicates the weight of local distortion; Indicates the basic distortion weight; λ represents the pattern sensitivity coefficient; λ represents the sensitivity enhancement adjustment coefficient, which ranges from 0.2 to 0.5, preferably 0.3; min represents the smaller value between the value in parentheses and 1. Therefore, with the same basic distortion weight, regions containing QR codes, barcodes, fine line textures, and continuous lines across corners can obtain higher local distortion weights.

[0152] S6: Based on the local distortion weight field, a reverse pre-distortion vector field that continuously varies along the attachment direction and attenuates in sections along the width direction is generated. Local reverse resampling is then performed on the original label pattern, outputting a pre-distortion label pattern that only affects the dynamic transition influence zone. In some embodiments, the implementation steps include:

[0153] Read each grid point G ij The corresponding local distortion weight W ij Local strain estimator Dynamic migration amount Smooth dynamic rotation center coordinates and grid point G ij The signed distance DU is relative to the u-direction of the dynamic angle broken line. ij Calculate grid point G ij Attachment deformation prediction vector The attached deformation prediction vector Including predicted deformation in the u direction and the predicted deformation in the v direction Thus, the predicted vector field of the attachment deformation is obtained. :

[0154] Among them, the predicted deformation in the u direction The calculation method is as follows: ;

[0155] Predicted deformation in the v direction The calculation formula is: ;

[0156] It should be noted that, This represents the coefficient representing the influence of local strain on the pattern offset in the u-direction; This represents the coefficient representing the influence of the bending center migration on the pattern offset in the u-direction; This represents the influence coefficient of the dynamic corner polyline change along the width direction on the pattern offset in the v direction; , and The calibration results can be obtained through calibration test patches using similar load-bearing objects, the same batch of label materials, and the same application process. The independent variables are the estimated local strain, dynamic migration, and the rate of change in the width of the dynamic corner break line in the calibration test patch sample. The dependent variables are the measured pattern offsets in the u-direction and v-direction. Least squares fitting is then used to obtain the final result. , and .

[0157] Based on local distortion weights Attach deformation prediction vector Convert to inverse predistortion vector ; For the inverse predistortion vector at discrete grid points Interpolation is performed to obtain the continuous inverse pre-distortion vector field R(u, v) covering the interior of the label design outline.

[0158] Specifically, the conversion formula is: ;

[0159] Specifically, for any point inside the label design outline = (u, v), read the inverse predistortion vectors of its surrounding adjacent grid points. And spline interpolation is used to calculate the inverse predistortion vector corresponding to this point. .in, This includes the reverse pre-distortion variable Rᵘ(u, v) in the u direction and the reverse pre-distortion variable Rᵛ(u, v) in the v direction;

[0160] For every point inside the Ω outline of the label design , = (u, v), and calculate its pre-distorted coordinates. After updating the coordinates of each point inside the label design outline Ω, a pre-distorted label pattern in vector form is regenerated.

[0161] Specifically, ;in, This represents the coordinates of the vector control points after pre-distortion; Represents the coordinates of the original vector control points; This represents the reverse predistortion vector at the control point.

[0162] Keeping the label design outline Ω as the final cutting boundary unchanged, only performing the above coordinate update or pixel resampling on the pattern points inside the label design outline Ω, and outputting the pre-distorted label pattern.

[0163] That is, the local distortion weight of the grid points outside the dynamic transition influence zone is reset to zero, so that the coordinates of the pattern outside the dynamic transition influence zone remain unchanged.

[0164] In this embodiment, when constructing the local distortion weight field, the basic distortion weight is fused with the pattern sensitivity coefficient, assigning higher local distortion weights to sensitive pattern elements such as QR codes, barcodes, fine line textures, and continuous lines across corners. Therefore, under the same angular deformation conditions, pattern areas with high requirements for recognizability and continuity can be given focused protection, improving the recognition reliability of QR codes and barcodes and reducing the risks of fine line texture breakage, misalignment of lines across corners, and discontinuities in pattern boundaries.

[0165] Furthermore, an inverse pre-distortion vector field is generated based on the local distortion weight field, and local inverse resampling is performed on the original label pattern. This makes the pre-distortion variable change continuously along the label attachment direction and attenuate in sections along the label width direction. This method enables the pre-distortion pattern to offset the local stretching, compression and offset deformation generated in the corner area after actual attachment, and realizes differentiated processing of the corner core area, transition area and plane preservation area, thereby improving the pattern integrity, recognition stability and appearance consistency of the label after it is attached across corners.

[0166] The foregoing has shown and described the basic principles, main features, and advantages of this application. Those skilled in the art should understand that this application is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of this application. Various changes and modifications can be made to this application without departing from the spirit and scope thereof, and all such changes and modifications fall within the scope of this application as claimed. The scope of protection of this application is defined by the appended claims and their equivalents.

Claims

1. A method for generating local pre-distortion of label patterns for corner areas, characterized in that, include: Obtain the 3D contour surface of the label carrier object and the label design data, establish a curvature analysis mesh, map each mesh point in the curvature analysis mesh to the 3D contour surface of the carrier object to obtain the corresponding surface points, and sort the surface points at the same width position along the label attachment direction to construct a width sampling line with normal, tangential and arc length positions. Based on the surface points, the changes in tangential angle, normal angle, and arc length deviation of adjacent points are calculated to generate a curvature change intensity sequence. Candidate corner regions are then separated based on the curvature change intensity sequence. By combining the candidate corner center of the candidate corner area and the attachment process parameters, the dynamic migration amount of the predicted bending center relative to the candidate corner center is calculated, and the dynamic corner line is determined based on the dynamic migration amount. Calculate the deformation influence of grid points, and search for transition boundaries that satisfy the preset cutoff conditions along each width sampling line from the dynamic corner broken line to both sides to generate a dynamic transition influence zone; Based on the dynamic corner polyline, the dynamic transition influence zone, and the label design outline, a local distortion weight field is constructed. Based on the local distortion weight field, an inverse predistortion vector field is generated, and based on the inverse predistortion vector field, local inverse resampling is performed on the original label pattern to output a predistorted label pattern that only affects the dynamic transition influence zone.

2. The method for generating a local pre-distortion of a label pattern facing a corner area according to claim 1, characterized in that, The method for obtaining the corresponding surface points includes: The u-direction is established based on the label attachment direction stored in the label design data, and the v-direction is established based on the label width direction stored in the label design data, forming a two-dimensional coordinate system for the label. Generate a curvature analysis mesh inside the label design outline according to the preset vertical sampling interval and the preset horizontal sampling interval; For each grid point, obtain the label center attachment line stored in the label design data, read the corresponding 3D path point, surface normal vector and path tangent vector on the label center attachment line, and calculate the 3D width direction vector. Based on the three-dimensional path point and the three-dimensional width direction vector, the initial three-dimensional corresponding point of the grid point is determined, and the initial three-dimensional corresponding point is projected onto the three-dimensional contour surface of the bearing object to obtain the corresponding surface point.

3. The method for generating a local pre-distortion of a label pattern for a corner area according to claim 2, characterized in that, The steps for constructing the width sampling line include: Surface points at the same width position are arranged in order along the u direction to form the j-th width sampling line. For each surface point in the width sampling line, the surface normal vector, local tangent vector, and arc length position are calculated, thus forming a width sampling line with normal, tangent, and arc length positions.

4. The method for generating a local pre-distortion of a label pattern facing a corner area according to claim 1, characterized in that, The method for generating curvature mutation intensity sequences includes: Calculate the tangential and normal changes for each surface point; Calculate the arc length deviation for each surface point; The curvature abrupt change intensity is obtained by weighted summation of the tangential change, normal change, and arc length deviation.

5. The method for generating a local pre-distortion of a label pattern facing a corner area according to claim 1, characterized in that, Methods for obtaining candidate corner regions based on curvature change intensity sequence separation include: Set a first curvature threshold and a second curvature threshold, wherein the first curvature threshold is less than the second curvature threshold; When the curvature abrupt change intensity is less than the first curvature threshold, the corresponding grid point is marked as a plane candidate point; When the curvature abrupt change intensity is greater than or equal to the first curvature threshold and less than or equal to the second curvature threshold, the corresponding grid point is marked as a transition candidate point; When the curvature change intensity is greater than the second curvature threshold, the corresponding grid point is marked as a corner candidate point, and consecutive adjacent corner candidate points are connected and merged to obtain a candidate corner region.

6. The method for generating a local pre-distortion of a label pattern facing a corner area according to claim 1, characterized in that, The method for calculating the dynamic migration amount includes: Read the curvature change intensity in the candidate corner region on each width sampling line, and determine the u-direction coordinate corresponding to the sampling point with the largest curvature change intensity as the candidate corner center; Obtain the bonding process parameters, including bonding tension, adhesive layer bonding strength and label material springback coefficient. Normalize the bonding process parameters to obtain the bending center migration coefficient. Calculate the width of the candidate corner region in the u-direction on the corresponding width sampling line; The dynamic migration amount is calculated based on the width in the U direction, the migration coefficient of the bending center, and the motion direction of the pressing mechanism stored in the label design data.

7. The method for generating a local pre-distortion of a label pattern facing a corner area according to claim 6, characterized in that, The method for determining the dynamic corner polyline includes: The dynamic corner center is calculated based on the candidate corner center and the dynamic migration amount. The coordinates of the dynamic corner center on adjacent width sampling lines are smoothed. The smoothed dynamic corner center coordinates are connected along the v direction to form a dynamic corner polyline.

8. The method for generating a local pre-distortion of a label pattern facing a corner area according to claim 7, characterized in that, The steps for determining the dynamic transition influence band include: For any grid point inside the label design outline, calculate the signed distance of that grid point relative to the u-direction of the dynamic corner polyline; Based on the arc length position corresponding to the grid point and the three-dimensional surface arc length increment between adjacent surface points, the three-dimensional surface arc length increment is compared with the design sampling distance in the label's two-dimensional coordinate system to calculate the local strain estimate. The deformation effect is calculated based on the curvature abrupt change intensity and the local strain estimate. Along each width sampling line, starting from the center of the smooth dynamic corner, search towards the negative side and the positive side of the u direction respectively. When the deformation influence of a consecutive preset number of sampling points is less than the preset transition cutoff threshold, determine the transition boundary point on the corresponding side, and form a dynamic transition influence zone based on the transition boundary points on both sides.

9. The method for generating a local pre-distortion of a label pattern facing a corner area according to claim 1, characterized in that, The steps for constructing the local distortion weight field include: Within the valid pattern area defined by the label design outline, read the pattern elements and mark QR codes, barcodes, fine line textures, and continuous lines across corners as pattern-sensitive elements; For each pattern-sensitive element, its element boundary, element type, and minimum feature width are recorded, and the element boundary is expanded outward based on the minimum feature width to form a sensitive influence area; The basic distortion weight of each grid point is calculated based on dynamic corner polylines and dynamic transition influence zones; The pattern sensitivity coefficient of each grid point is determined based on the aforementioned sensitive influence area; By fusing the basic distortion weights and the pattern sensitivity coefficients, the local distortion weights of each grid point are obtained, and the local distortion weight field is formed by the local distortion weights of all grid points inside the label design outline.

10. The method for generating a local pre-distortion of a label pattern for a corner area according to claim 1, characterized in that, The step of generating the inverse predistortion vector field includes: Read the local distortion weight, local strain estimate, dynamic migration, smoothed dynamic corner center coordinates, and the signed distance of the grid point relative to the u-direction of the dynamic corner polyline for each grid point; The predicted deformation in the u-direction is calculated based on the local strain estimate and the dynamic migration. The predicted deformation in the v direction is calculated based on the change in the dynamic corner line along the label width direction and the signed distance in the u direction. The predicted deformation in the u-direction and the predicted deformation in the v-direction constitute the attachment deformation prediction vector field. The attachment deformation prediction vector field is reverse-transformed according to the local distortion weight to obtain a discrete reverse pre-distortion vector. The discrete reverse pre-distortion vector is interpolated to generate a reverse pre-distortion vector field covering the inside of the label design outline.

Images