Aluminum veneer engraving path automatic correction method and system based on positioning image feedback

Abstract

This invention discloses an automatic correction method and system for aluminum single-panel engraving paths based on positioning image feedback, specifically relating to the field of digital program-controlled path correction technology. It addresses the problem that path correction methods often include protective film pseudo-edges in spatial transformation calculations, causing the corrected path to deviate from the true contour of the aluminum single-panel. The method involves acquiring a positioning image containing the state of the protective film on the aluminum single-panel surface and extracting boundary features. The material contour from the original tool path data is then matched with a predetermined cutting contour to determine candidate pseudo-features. Brightness gradient asymmetry is calculated along the normal direction of the boundary features to identify protective film tear pseudo-edges. The coordinates of process feature points are retrieved to verify the protective film tear pseudo-edges. After removing protective film tear pseudo-edges and candidate pseudo-features that maintain pseudo-edge determination, a reliable boundary feature set is constructed from the remaining boundary features. The spatial transformation relationship is calculated using only the reliable boundary feature set as a constraint, and the original tool path data is then transformed as a whole.

Description

Technical Field

[0001] This invention relates to the field of digital program control path correction technology, and more specifically, to an automatic correction method and system for aluminum single-panel engraving paths based on positioning image feedback. Background Technology

[0002] In CNC engraving of aluminum panels for curtain walls, the control system typically drives the cutting tool according to a pre-programmed machining procedure to complete processes such as grooving and cutting. Since the actual placement and dimensions of the aluminum panels after loading and clamping often deviate from the ideal state set in the program, the industry can introduce an automatic correction method based on visual feedback. This method uses an imaging device installed on the engraving machine to acquire a positioning image of the aluminum panel. By extracting the panel boundary features from the image, the spatial transformation relationship of the actual panel relative to the program-set reference is calculated. This allows for a comprehensive transformation of the tool path data in the original program, generating a corrected machining path. This process is entirely completed within the CNC system's program data processing stage and does not involve adjustments to the machine tool's mechanical structure.

[0003] However, the boundary features extracted from the positioning image of the aluminum panel in the image processing stage do not always accurately reflect the contour of the aluminum panel itself. Affected by the state of the protective film on the surface of the aluminum panel, the protective film may experience local delamination, bulging, edge warping, or partial tearing during handling or clamping. These film abnormalities will form pseudo edges that are parallel or intersecting with the real panel edges during imaging. Without identifying the source of the features, the existing path correction logic includes these pseudo edges in the calculation of the spatial transformation model along with the real panel edge features, resulting in the generated path correction data containing unpredictable distortion components. Ultimately, this causes the corrected processing path to deviate from the real processing contour of the aluminum panel. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, the present invention provides an automatic correction method and system for aluminum single-panel engraving path based on positioning image feedback to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] The automatic correction method for engraving paths on aluminum single-panel panels based on positioning image feedback includes the following steps:

[0007] S1: Obtain a positioning image containing the state of the protective film on the surface of the aluminum panel and extract multiple boundary features;

[0008] S2: Retrieve the sheet metal contour and the predetermined cutting contour from the original toolpath data, match each boundary feature with the sheet metal contour and the predetermined cutting contour, and determine the boundary features that cannot be matched with the corresponding contour as candidate pseudo features.

[0009] S3: Extract the brightness distribution along the normal direction for each boundary feature and calculate the brightness gradient asymmetry. Identify those with brightness gradient asymmetry below the preset symmetry threshold as protective film tear pseudo-edges superimposed on the real boundary.

[0010] S4: Retrieve the coordinates of process feature points in the original toolpath data and verify the false edges of the protective film tear. If the geometric extension path crosses or aligns with the coordinates of the process feature points, the false edge judgment is maintained.

[0011] S5: Remove the protective membrane tear pseudo-edge and candidate pseudo-features that maintain the pseudo-edge determination after verification from multiple boundary features, and form a credible boundary feature set from the remaining boundary features.

[0012] S6: Calculate the spatial transformation relationship using only the credible boundary feature set as a constraint, and perform an overall transformation on the original toolpath data based on the spatial transformation relationship to generate the corrected machining path.

[0013] Furthermore, a positioning image containing the state of the protective film on the surface of the aluminum single panel is acquired, and multiple boundary features are extracted, including: acquiring a positioning image containing the state of the protective film on the surface of the aluminum single panel, identifying pixels with abrupt changes in brightness in the positioning image, connecting the pixels with abrupt changes in brightness to form multiple continuous lines, and extracting the continuous lines that are closed or cross the boundary of the positioning image as boundary features.

[0014] Furthermore, the sheet metal contour and the predetermined cutting contour in the original toolpath data are retrieved, and each boundary feature is matched with the sheet metal contour and the predetermined cutting contour. Boundary features that cannot be matched with the corresponding contour are judged as candidate pseudo features. This includes: retrieving the sheet metal contour and the predetermined cutting contour defined in the original toolpath data, converting the sheet metal contour into a first polygon representation, converting the predetermined cutting contour into a second polygon representation, converting each boundary feature into a third polygon representation, calculating the shape similarity between the third polygon representation and the first polygon representation and the shape similarity between the third polygon representation and the second polygon representation, and judging the boundary features whose shape similarity is lower than the preset matching threshold as candidate pseudo features.

[0015] Further, the brightness distribution of each boundary feature is extracted along the normal direction and the brightness gradient asymmetry is calculated. The brightness gradient asymmetry below a preset symmetry threshold is identified as a protective film tear pseudo-edge superimposed on the real boundary. This includes: taking the determined candidate pseudo-features as processing objects, taking multiple sampling points along the extension direction of the candidate pseudo-features at a preset step size, extracting the brightness value sequence along the normal direction of the candidate pseudo-features at each sampling point, calculating the ratio of the gradient magnitude of the brightness falling edge to the gradient magnitude of the brightness rising edge in the brightness value sequence, using the ratio as the brightness gradient asymmetry, and identifying the candidate pseudo-features with brightness gradient asymmetry below a preset symmetry threshold as a protective film tear pseudo-edge superimposed on the real boundary.

[0016] Further, the ratio of the gradient magnitude of the falling edge to the gradient magnitude of the rising edge in the brightness value sequence is calculated, including: at each sampling point of the candidate pseudo-feature, traversing the brightness value sequence from one side of the candidate pseudo-feature to the other along the normal direction, identifying the continuous pixel segment where the brightness value transitions from the high brightness range to the low brightness range as the falling edge and recording its gradient magnitude, identifying the continuous pixel segment where the brightness value transitions from the low brightness range to the high brightness range as the rising edge and recording its gradient magnitude, and dividing the gradient magnitude of the falling edge by the gradient magnitude of the rising edge to obtain the brightness gradient asymmetry.

[0017] Furthermore, the coordinates of process feature points in the original toolpath data are retrieved to verify the pseudo-edge of the protective film tear. If its geometric extension path crosses or aligns with the coordinates of the process feature points, the pseudo-edge determination is maintained. This includes: retrieving the coordinates of the chamfer vertex and the contour tip recorded in the original toolpath data as the coordinates of the process feature points, generating a geometric extension path along the extension direction of the pseudo-edge of the protective film tear, calculating the minimum distance between the geometric extension path and the coordinates of the process feature points, and determining the pseudo-edge of the protective film tear that the minimum distance is lower than a preset distance threshold as crossing or aligning with the coordinates of the process feature points, thus maintaining the pseudo-edge determination for the pseudo-edge of the protective film tear.

[0018] Further, generating a geometric extension path along the extension direction of the protective film tear pseudo edge includes: obtaining the two endpoints of the protective film tear pseudo edge, extending outward from the two endpoints along the tangent direction of the protective film tear pseudo edge at the endpoints respectively, the extension length being a preset proportion of the total length of the protective film tear pseudo edge, and connecting the protective film tear pseudo edge and its two end extension segments to form a geometric extension path.

[0019] Furthermore, protective membrane tear pseudo-edges and candidate pseudo-features that maintain pseudo-edge determination after verification are removed from multiple boundary features, and the remaining boundary features constitute a credible boundary feature set, including: removing protective membrane tear pseudo-edges that maintain pseudo-edge determination from multiple boundary features, removing boundary features that are determined to be candidate pseudo-features from multiple boundary features, and taking all the remaining boundary features after removal as the credible boundary feature set.

[0020] Furthermore, the spatial transformation relationship is calculated using only the credible boundary feature set as a constraint. Based on the spatial transformation relationship, the original toolpath data is transformed as a whole to generate the corrected machining path. This includes: using the boundary features contained in the credible boundary feature set as spatial reference points, calculating the rigid body transformation matrix that minimizes the positional deviation between the credible boundary feature set and the corresponding contours in the original toolpath data, using the rigid body transformation matrix as the spatial transformation relationship, and applying the spatial transformation relationship to the coordinates of all toolpath points in the original toolpath data to generate the corrected machining path.

[0021] On the other hand, the present invention provides an automatic correction system for engraving paths on aluminum single-panel panels based on positioning image feedback, comprising the following modules:

[0022] The image acquisition module is used to acquire positioning images containing the state of the protective film on the surface of aluminum panels and extract multiple boundary features;

[0023] The feature matching module is used to retrieve the sheet metal contour and the predetermined cutting contour from the original tool path data, match each boundary feature with the sheet metal contour and the predetermined cutting contour, and determine the boundary features that cannot be matched with the corresponding contour as candidate pseudo features.

[0024] The brightness analysis module is used to extract the brightness distribution along the normal direction of each boundary feature and calculate the brightness gradient asymmetry. The brightness gradient asymmetry is lower than the preset symmetry threshold and is identified as a protective film tear pseudo edge superimposed on the real boundary.

[0025] The geometric verification module is used to retrieve the coordinates of process feature points in the original toolpath data and verify the false edges of the protective film tear. If its geometric extension path crosses or aligns with the coordinates of the process feature points, the false edge judgment is maintained.

[0026] The feature filtering module is used to remove protective membrane tear pseudo-edges and candidate pseudo-features from multiple boundary features after verification to maintain the pseudo-edge determination, and to form a credible boundary feature set from the remaining boundary features.

[0027] The path generation module is used to calculate the spatial transformation relationship based solely on the set of credible boundary features, and to perform an overall transformation on the original toolpath data based on the spatial transformation relationship to generate the corrected machining path.

[0028] Compared with the prior art, the present invention has the following beneficial effects:

[0029] 1. When automatically correcting the engraving path through positioning image feedback, the extracted boundary features are reliably identified in multiple dimensions. During the CNC system's program data processing, the boundary features are first matched and judged using the sheet metal contour and the predetermined cutting contour in the original tool path data, and candidate pseudo features that cannot correspond to the program-defined contour are screened out. Then, the brightness gradient asymmetry is analyzed along the normal direction of the boundary features to identify the pseudo edges superimposed on the real boundary due to the tearing of the protective film. Finally, the identification results are verified by the coordinates of the process feature points, forming a multi-step pseudo edge screening chain. This effectively separates the pseudo edges caused by the delamination, bulging, warping, or tearing of the protective film from the boundary features used for path correction, avoiding the introduction of distortion components by including these pseudo edges in the spatial transformation model calculation as is. This ensures that the corrected processing path accurately corresponds to the real processing contour of the aluminum single panel body.

[0030] 2. The constraints of the entire correction process are derived from the boundary feature set that has been reliably verified. When calculating the spatial transformation relationship, it only relies on the reliable boundary features that reflect the true boundary of the aluminum panel body, and is no longer affected by the false edges of the protective film. All data processing from image acquisition, feature verification to path correction is completed within the CNC system. It does not rely on the adjustment of the machine tool's mechanical structure. The result of path correction is directly reflected in the overall transformation of the tool path data, which enables the aluminum panel engraving process to obtain an accurate processing path even when the protective film has defects, avoiding path deviation caused by abnormal protective film conditions. Attached Figure Description

[0031] Figure 1 This is a flowchart of the automatic correction method for engraving path of aluminum single plate based on positioning image feedback according to the present invention;

[0032] Figure 2 This is a schematic diagram of the automatic correction system for aluminum single-panel engraving paths based on positioning image feedback according to the present invention. Detailed Implementation

[0033] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0034] Example 1: Figure 1 The present invention provides an automatic correction method for engraving paths on aluminum single-panel panels based on positioning image feedback, which includes the following steps:

[0035] S1: Obtain a positioning image containing the state of the protective film on the surface of the aluminum panel and extract multiple boundary features;

[0036] S2: Retrieve the sheet metal contour and the predetermined cutting contour from the original toolpath data, match each boundary feature with the sheet metal contour and the predetermined cutting contour, and determine the boundary features that cannot be matched with the corresponding contour as candidate pseudo features.

[0037] S3: Extract the brightness distribution along the normal direction for each boundary feature and calculate the brightness gradient asymmetry. Identify those with brightness gradient asymmetry below the preset symmetry threshold as protective film tear pseudo-edges superimposed on the real boundary.

[0038] S4: Retrieve the coordinates of process feature points in the original toolpath data and verify the false edges of the protective film tear. If the geometric extension path crosses or aligns with the coordinates of the process feature points, the false edge judgment is maintained.

[0039] S5: Remove the protective membrane tear pseudo-edge and candidate pseudo-features that maintain the pseudo-edge determination after verification from multiple boundary features, and form a credible boundary feature set from the remaining boundary features.

[0040] S6: Calculate the spatial transformation relationship using only the credible boundary feature set as a constraint, and perform an overall transformation on the original toolpath data based on the spatial transformation relationship to generate the corrected machining path.

[0041] In step S1, a positioning image containing the state of the protective film on the surface of the aluminum single panel is acquired. Pixels with abrupt changes in brightness are identified in the positioning image. Pixels with abrupt changes in brightness are connected to form multiple continuous lines. Continuous lines that are closed or cross the boundary of the positioning image are extracted as boundary features.

[0042] The imaging device of the engraving machine is mounted on the side of the spindle, with its optical axis perpendicular to the worktable plane. After the aluminum panel is placed on the worktable and clamped, the imaging device photographs the surface of the aluminum panel to obtain a positioning image that includes the state of the protective film. In the imaging parameters, the exposure time is set to a value that makes the texture of the protective film surface discernible in the positioning image. The exposure time is determined by pre-taking a sample of the aluminum panel with the protective film and selecting an exposure time value that creates a grayscale difference between the frosted texture of the protective film and the coating texture of the aluminum panel in the positioning image. For example, the exposure time can be set to a value within the range of 50 microseconds to 500 microseconds.

[0043] After obtaining the localization image, pixels with abrupt brightness changes are identified. A difference method is used to calculate the brightness difference between adjacent pixels along both the row and column directions of the localization image. Pixels with a brightness difference exceeding a preset row-direction brightness difference threshold are marked as horizontal brightness change pixels, and pixels with a brightness difference exceeding a preset column-direction brightness difference threshold are marked as vertical brightness change pixels. The union of the horizontal and vertical brightness change pixels is taken as the pixel where the brightness change occurs. The preset row-direction brightness difference threshold and preset column-direction brightness difference threshold are set as follows: Select a region in the positioning image that contains only the surface texture of the protective film and has no aluminum plate edges or protective film defects as a reference background region. Calculate the average value of the brightness difference of all adjacent pixels in the reference background region. Set the preset row-direction brightness difference threshold to several times the average value. For example, multiply the average value by a value between 2.5 and 4.0 to obtain the preset row-direction brightness difference threshold. Set the preset column-direction brightness difference threshold to several times the average value. For example, multiply the average value by a value between 2.5 and 4.0 to obtain the preset column-direction brightness difference threshold. The preset row-direction brightness difference threshold and the preset column-direction brightness difference threshold can be the same value, or different values ​​can be taken according to the difference in the thickness of the protective film texture in the row direction and the column direction.

[0044] After identifying pixels with abrupt changes in brightness, these pixels are connected to form multiple continuous lines. The connection process is as follows: Scanning row by row in the positioning image, adjacent pixels with abrupt changes in brightness within the same row, with a spacing not exceeding a preset discontinuity, are grouped into the same row-direction continuous line segment. Scanning column by column in the positioning image, adjacent pixels with abrupt changes in brightness within the same column, with a spacing not exceeding the preset discontinuity, are grouped into the same column-direction continuous line segment. The row-direction continuous line segments and column-direction continuous line segments are merged. At the intersection of the row-direction continuous line segments and column-direction continuous line segments, they are connected to form a continuous line network covering the entire positioning image. The preset discontinuity distance is determined based on the spatial resolution of the imaging device. When the pixel pitch of the imaging device is 0.05mm to 0.15mm, the preset discontinuity distance is set to 2 to 5 pixels wide, ensuring that locally missing pixels with abrupt changes in brightness caused by slight discontinuities in the protective film surface texture or imaging noise are connected without affecting the edge orientation.

[0045] After obtaining a continuous line network covering the entire positioning image, closed continuous lines and continuous lines crossing the positioning image boundary are identified from the continuous line network. Closed continuous lines are identified as follows: starting from any node in the continuous line network, tracing along the continuous line; if the tracing path returns to the starting point and does not contain repeating continuous line segments, then all continuous line segments traversed by the tracing path are considered closed continuous lines. Continuous lines crossing the positioning image boundary are identified as follows: examining the endpoints of each continuous line in the continuous line network; if one endpoint of the continuous line is located on a boundary pixel row or column of the positioning image, and the other endpoint of the continuous line is also located on a boundary pixel row or column of the positioning image, or if one endpoint of the continuous line is located on a boundary pixel row or column of the positioning image while the other endpoint terminates inside the continuous line network, then the continuous line is considered to cross the positioning image boundary. During the identification process, for continuous lines within a continuous line network that are neither closed nor have their endpoints touching the boundary of the positioning image, if the length of the continuous line exceeds a preset connectivity length threshold, it is merged with adjacent closed continuous lines or continuous lines crossing the boundary of the positioning image. If the length of the continuous line does not exceed the preset connectivity length threshold, it is discarded as an isolated line segment. The preset connectivity length threshold is set by obtaining the length values ​​of all continuous lines in the continuous line network that are neither closed nor have their endpoints touching the boundary of the positioning image, and setting the preset connectivity length threshold to a multiple of the median of these length values, for example, multiplying the median by a value between 0.3 and 0.7.

[0046] All identified closed continuous lines and continuous lines crossing the boundaries of the localization image are extracted as boundary features. Each boundary feature is stored as a sequence of pixel coordinates on the continuous line, while retaining the row coordinates, column coordinates, and original brightness value of each pixel in the localization image.

[0047] In step S2, the sheet metal contour and the predetermined cutting contour defined in the original tool path data are retrieved. The sheet metal contour is converted into a first polygon representation, the predetermined cutting contour is converted into a second polygon representation, and each boundary feature is converted into a third polygon representation. The shape similarity between the third polygon representation and the first polygon representation and the shape similarity between the third polygon representation and the second polygon representation are calculated respectively. Boundary features whose shape similarity is lower than the preset matching threshold are judged as candidate pseudo features.

[0048] The original toolpath data is stored in the CNC device of the engraving machine. This data includes the sheet metal contour definition and the predetermined cutting contour definition. The sheet metal contour is the ideal outer contour of the sheet metal to be processed, set during programming. The predetermined cutting contour is the projection contour of the tool cutting trajectory planned during programming onto the sheet metal plane. The contour point coordinate sequence of the sheet metal contour definition is retrieved from the CNC device, and the sequence is connected end-to-end in adjacent order to form a first polygon. The contour point coordinate sequence of the predetermined cutting contour definition is retrieved from the CNC device, and the sequence is connected end-to-end in adjacent order to form a second polygon. For each boundary feature extracted in step S1, the pixel coordinate sequence stored for that boundary feature is read, and the sequence is connected end-to-end in adjacent order to form a third polygon. When the pixel coordinate sequence of the boundary feature corresponds to a closed continuous line, the third polygon is a closed polygon. When the pixel coordinate sequence of the boundary feature corresponds to a continuous line crossing the boundary of the positioning image, the virtual connection between the two endpoints crossing the boundary of the positioning image is connected to the pixel coordinate sequence to form a closed third polygon.

[0049] After obtaining the first, second, and third polygon representations, the shape similarity between the third and first polygon representations, and between the third and second polygon representations, is calculated. The shape similarity is calculated as follows: The first, second, and third polygon representations are placed in the same coordinate system. Polygon simplification is performed on each of the three representations by sampling along the polygon boundaries at fixed arc length intervals. The fixed arc length interval is set as a preset proportion of the minimum total length of the boundaries of each polygon representation, for example, a value between 1 / 20 and 1 / 50. After sampling, sample point sets for the first, second, and third polygon representations are obtained. For each sample point in the sample point set of the third polygon representation, the Euclidean distance from that sample point to the nearest sample point in the sample point set of the first polygon representation is calculated. The average of all Euclidean distances is used as the shape difference measure between the third and first polygon representations, and the shape similarity is the reciprocal of this shape difference measure. The shape similarity between the third polygon representation and the second polygon representation is calculated as follows: For each sampling point in the sampling point set of the third polygon representation, calculate the Euclidean distance from the sampling point to the nearest sampling point in the sampling point set of the second polygon representation. The average value of all Euclidean distances is used as the shape difference measure between the third polygon representation and the second polygon representation, and the shape similarity is taken as the reciprocal of the shape difference measure.

[0050] The preset matching threshold is set as follows: After determining the outline of the aluminum panel in the current batch and the predetermined cutting outline, a portion of the multiple boundary features extracted in step S1 are selected as calibration samples. The operator marks the boundary features that clearly belong to the real boundary of the aluminum panel and the boundary features that clearly belong to the protective film defects in the calibration samples. The shape similarity between each marked boundary feature and the shape similarity between the first polygon and the second polygon is calculated. The boundary value between the shape similarity value corresponding to the boundary feature that clearly belongs to the real boundary of the aluminum panel and the shape similarity value corresponding to the boundary feature that clearly belongs to the protective film defects is taken as the preset matching threshold. For example, the minimum value of the shape similarity value corresponding to all the boundary features that clearly belong to the real boundary of the aluminum panel and the maximum value of the shape similarity value corresponding to all the boundary features that clearly belong to the protective film defects are divided by 2 to obtain the preset matching threshold.

[0051] After calculating the shape similarity between the third polygon representation and the first polygon representation, and the shape similarity between the third polygon representation and the second polygon representation, the two shape similarities are compared with preset matching thresholds. When the shape similarity between the third polygon representation and the first polygon representation is lower than the preset matching threshold, and the shape similarity between the third polygon representation and the second polygon representation is also lower than the preset matching threshold, the boundary feature corresponding to the third polygon representation is determined as a candidate false feature. A candidate false feature indicates that the boundary feature does not correspond to the ideal outer contour of the aluminum single-panel defined by the sheet metal contour, nor to the tool cutting trajectory projection contour defined by the predetermined cutting contour, and has a high probability of being a false edge generated by an abnormal protective film.

[0052] When implementing step S3, the candidate pseudo-features determined in step S2 are used as processing objects. Multiple sampling points are taken along the extension direction of the candidate pseudo-features at a preset step size. At each sampling point, a brightness value sequence is extracted along the normal direction of the candidate pseudo-features. The ratio of the gradient magnitude of the brightness falling edge to the gradient magnitude of the brightness rising edge in the brightness value sequence is calculated. The ratio is used as the brightness gradient asymmetry. Candidate pseudo-features with brightness gradient asymmetry lower than a preset symmetry threshold are identified as protective film tear pseudo-edges superimposed on the real boundary.

[0053] After candidate pseudo-features are identified in step S2, each candidate pseudo-feature is processed separately. Candidate pseudo-features are stored as a sequence of pixel coordinates, containing the row coordinates, column coordinates, and original brightness value of each pixel. Multiple sampling points are taken along the extension direction of the candidate pseudo-feature at a preset step size. The preset step size is set as follows: the total length of the pixel coordinate sequence of the candidate pseudo-feature is obtained, and the preset step size is set to a fixed proportion of this total length. For example, the preset step size can be set to a value between 1 / 50 and 1 / 20 of the total length, or it can be set to an integer multiple of the pixel pitch of the imaging device. For example, when the pixel pitch is 0.05mm to 0.15mm, the preset step size is 5 to 15 pixels wide. Sampling points are taken starting from one endpoint of the candidate pseudo-feature and taking one sampling point every preset step size along the extension direction of the candidate pseudo-feature until the entire length of the candidate pseudo-feature is covered. At each sampling point, the normal direction of the candidate pseudo-feature at that sampling point is determined, and the normal direction is perpendicular to the extension direction of the candidate pseudo-feature at that sampling point.

[0054] At each sampling point, a brightness value sequence is extracted along the normal direction of the candidate pseudo-feature. The extraction method is as follows: starting from the sampling point, brightness values ​​are read pixel by pixel along the normal direction towards one side of the candidate pseudo-feature, with a reading length of the first normal sampling length. Then, starting from the sampling point, brightness values ​​are read pixel by pixel along the normal direction towards the other side of the candidate pseudo-feature, with a reading length of the second normal sampling length. The brightness values ​​read from both sides are then concatenated in order from one side to the other to form a brightness value sequence. The first and second normal sampling lengths are taken to have the same value, for example, both the first and second normal sampling lengths are 15 to 40 pixels wide. This value is more than twice the pixel size of the protective film thickness, so that the brightness value sequence can completely cover the brightness change area on both sides of the pseudo edge formed by the tear in the protective film.

[0055] After obtaining the brightness value sequence at each sampling point, at each sampling point of the candidate pseudo-feature, the brightness value sequence is traversed from one side of the candidate pseudo-feature to the other along the normal direction. Continuous pixel segments where the brightness value transitions from a high-brightness range to a low-brightness range are identified as falling edges, and their gradient magnitudes are recorded. Continuous pixel segments where the brightness value transitions from a low-brightness range to a high-brightness range are identified as rising edges, and their gradient magnitudes are recorded. The brightness gradient asymmetry is obtained by dividing the gradient magnitude of the falling edge by the gradient magnitude of the rising edge. The falling edge is identified by traversing the brightness difference between adjacent pixels in the brightness value sequence. When the brightness values ​​of multiple consecutive pixels show a monotonically decreasing trend and the difference between the brightness values ​​of the first and last pixels exceeds a preset falling edge brightness difference threshold, the continuous pixel segment is identified as a falling edge. The method for identifying the rising edge of brightness is to iterate through the brightness difference between adjacent pixels in the brightness value sequence. When the brightness values ​​of multiple consecutive pixels show a monotonically increasing trend and the difference between the brightness value of the last pixel and the first pixel exceeds a preset rising edge brightness difference threshold, the consecutive pixel segment is identified as a rising edge of brightness. The preset falling edge brightness difference threshold and the preset rising edge brightness difference threshold are set as follows: In the positioning image, a region containing only the surface texture of the protective film and without aluminum plate edges or protective film defects is selected as a reference background region. The maximum value of the brightness difference between adjacent pixels in the reference background region is calculated, and the preset falling edge brightness difference threshold is set to several times this maximum value, for example, multiplying the maximum value by a value between 1.5 and 3.0. The preset rising edge brightness difference threshold is also set to several times this maximum value, for example, multiplying the maximum value by a value between 1.5 and 3.0.

[0056] The gradient magnitude of the falling edge of brightness is recorded by dividing the difference between the brightness value of the first pixel and the brightness value of the last pixel in the falling edge by the number of pixels spanned by the falling edge. Similarly, the gradient magnitude of the rising edge of brightness is recorded by dividing the difference between the brightness value of the last pixel and the brightness value of the first pixel in the rising edge by the number of pixels spanned by the rising edge. The ratio of the gradient magnitude of the falling edge to the gradient magnitude of the rising edge at each sampling point is taken as the brightness gradient asymmetry at that sampling point. The average of the brightness gradient asymmetries of all sampling points on the candidate pseudo-feature is taken as the brightness gradient asymmetry of the candidate pseudo-feature.

[0057] The preset symmetry threshold is set as follows: After determining the positioning image of the current batch of aluminum panels, select the boundary features that clearly belong to the real boundary of the aluminum panel from the multiple boundary features extracted in step S1 as calibration samples. For the calibration samples, the operation of extracting the brightness value sequence along the normal direction and calculating the brightness gradient asymmetry is also performed. The brightness gradient asymmetry value of the calibration sample is used as the reference value of the brightness gradient asymmetry of the real boundary. The sum of the minimum and maximum values ​​of the reference values ​​of the brightness gradient asymmetry of the real boundary is divided by 2 to obtain the preset symmetry threshold. Alternatively, the preset symmetry threshold can be set to 1.0, so that when the gradient amplitude of the brightness falling edge is equal to or close to the gradient amplitude of the brightness rising edge, it is symmetric, and when the gradient amplitude of the brightness falling edge is significantly smaller than the gradient amplitude of the brightness rising edge, it is asymmetric.

[0058] Candidate pseudo-features with brightness gradient asymmetry below a preset symmetry threshold are identified as protective film tear pseudo-edges superimposed on the real boundary. When the brightness gradient asymmetry of a candidate pseudo-feature is below the preset symmetry threshold, it indicates that the gradient magnitude of the brightness falling edge in the normal direction is significantly smaller than the gradient magnitude of the brightness rising edge, and the brightness distribution exhibits asymmetrical characteristics on both sides. This is precisely the imaging characteristic caused by the same material on both sides of the edge formed by the protective film tear, and the candidate pseudo-feature is identified as a protective film tear pseudo-edge superimposed on the real boundary. When the brightness gradient asymmetry of a candidate pseudo-feature is not lower than the preset symmetry threshold, it indicates that the gradient magnitude of the brightness falling edge in the normal direction is close to the gradient magnitude of the brightness rising edge, and the brightness distribution exhibits symmetrical characteristics on both sides, which is consistent with the imaging characteristics of different materials on both sides of the real boundary of the aluminum panel. This candidate pseudo-feature is not identified as a protective film tear pseudo-edge.

[0059] When implementing step S4, the coordinates of the chamfer vertex and the contour tip recorded in the original tool path data are retrieved as the coordinates of the process feature point. A geometric extension path is generated along the extension direction of the protective film tear pseudo-edge. The minimum distance between the geometric extension path and the coordinates of the process feature point is calculated. The protective film tear pseudo-edge with a minimum distance lower than the preset distance threshold is determined to cross or align with the coordinates of the process feature point. The pseudo-edge determination of the protective film tear pseudo-edge is maintained.

[0060] After identifying the false edges of the protective film tear in step S3, each false edge is individually verified. The original toolpath data is retrieved from the CNC device. This data defines the geometric information of the tool cutting trajectory, which includes trajectory segments corresponding to the corner cutting process and the contour sharp corner process. The trajectory segment corresponding to the corner cutting process consists of two intersecting straight line segments formed when the tool cuts at the corner of the sheet metal; the coordinates of the intersection of these two lines are the coordinates of the corner vertex. The trajectory segment corresponding to the contour sharp corner process is the turning point where the tool's direction abruptly changes during cutting along the sheet metal contour; the coordinates of this turning point are the coordinates of the contour sharp point. All trajectory segments in the original toolpath data are traversed, and all corner vertex coordinates and all contour sharp point coordinates are extracted and aggregated into a process feature point coordinate set. Each process feature point coordinate in the set is stored in two-dimensional coordinate form.

[0061] A geometric extension path is generated along the extension direction of the protective film tear pseudo-edge. The generation process is as follows: The two endpoints of the protective film tear pseudo-edge are obtained. The endpoints of the protective film tear pseudo-edge refer to the two end pixels of the continuous line corresponding to the protective film tear pseudo-edge in the boundary features extracted in step S1. Starting from the first endpoint, the tangent direction of the protective film tear pseudo-edge at the first endpoint is calculated. The tangent direction is calculated by taking several pixels on the protective film tear pseudo-edge closest to the first endpoint, for example, 3 to 8 pixels, and fitting a straight line to these pixels. The direction of the fitted straight line is taken as the tangent direction of the protective film tear pseudo-edge at the first endpoint. Extending outward from the first endpoint along this tangent direction, the extension length is a preset proportion of the total length of the protective film tear pseudo-edge, for example, the extension length is the total length of the protective film tear pseudo-edge multiplied by a value between 1 / 8 and 1 / 3, to obtain the first extension segment. Starting from the second endpoint, calculate the tangent direction of the protective film tear pseudo-edge at the second endpoint. The tangent direction is calculated by taking several pixels on the protective film tear pseudo-edge closest to the second endpoint, for example, 3 to 8 pixels, and fitting a straight line to these pixels. The direction of the fitted line is taken as the tangent direction of the protective film tear pseudo-edge at the second endpoint. Extend outward from the second endpoint along this tangent direction. The extension length is a preset proportion of the total length of the protective film tear pseudo-edge, for example, the extension length is a value between 1 / 8 and 1 / 3 of the total length of the protective film tear pseudo-edge, to obtain the second extension segment. Connect the first extension segment, the protective film tear pseudo-edge itself, and the second extension segment end to end to form a geometric extension path.

[0062] The minimum distance between the geometric extension path and the coordinates of the process feature points is calculated as follows: The geometric extension path is discretized into multiple path points, with the spacing between the discretization points being 1 to 2 times the pixel pitch of the imaging device. For example, when the pixel pitch is 0.05 mm to 0.15 mm, the spacing between the discretization points is 0.05 mm to 0.30 mm. For each path point of the geometric extension path, the Euclidean distance from that path point to the coordinates of every process feature point in the set of process feature point coordinates is calculated. The minimum value among all Euclidean distances is taken as the minimum distance between the geometric extension path and the coordinates of the process feature points.

[0063] The preset distance threshold is set as follows: After determining the original tool path data of the current batch of aluminum single panels, select the boundary features that clearly belong to the real boundary of the aluminum single panel from the multiple boundary features extracted in step S1 as calibration samples. For the calibration samples, perform the same operation of generating geometric extension paths and calculating the minimum distance between the geometric extension paths and the coordinates of the process feature points. Use the minimum distance value corresponding to the calibration samples as the reference value of the process-related distance of the real boundary. Take the maximum value among all the reference values ​​of the process-related distance of the real boundary as the preset distance threshold, or set the preset distance threshold to a number of times the pixel spacing of the imaging device. For example, set the preset distance threshold to a value between 3 and 8 times the pixel spacing, so that the preset distance threshold can accommodate the positional deviation caused by the imaging resolution and the pixel coordinate extraction accuracy.

[0064] If the minimum distance between the protective film tear pseudo-edge and the process feature point coordinates is lower than the preset distance threshold, it is determined that the protective film tear pseudo-edge crosses or aligns with the coordinates of the process feature point, and the pseudo-edge determination is maintained. When the minimum distance between the geometric extension path and the coordinates of the process feature point is lower than the preset distance threshold, it indicates that the geometric extension path of the protective film tear pseudo-edge crosses or aligns with the corner vertex or contour tip defined in the original toolpath data in space. The formation of the protective film tear pseudo-edge is consistent with the physical law that the protective film is easy to tear at corners or sharp corners. The review is passed, and the pseudo-edge determination made in step S3 is maintained. When the minimum distance between the geometric extension path and the coordinates of the process feature point is not lower than the preset distance threshold, it indicates that the geometric extension path of the protective film tear pseudo-edge does not cross or align with the corner vertex and contour tip defined in the original toolpath data in space. The formation of the protective film tear pseudo-edge cannot be explained by the law that the protective film is easy to tear at process feature points. The review is not passed, and the protective film tear pseudo-edge is restored to a pending feature. Boundary features restored to pending features are left for subsequent judgment by the operator, or they are treated as non-protective film tear pseudo-edges in step S5 and not included in the scope of protective film tear pseudo-edges removed in step S5.

[0065] In step S5, the protective membrane tearing pseudo-edge that maintains the pseudo-edge determination is removed from multiple boundary features, and the boundary features that are determined to be candidate pseudo-features are removed from multiple boundary features. All the remaining boundary features after removal are taken as the credible boundary feature set.

[0066] After verifying the false edges of the protective film tear in step S4, the verification result for each false edge of the protective film tear is obtained. The verification results are divided into two categories: the first category is a false edge maintenance judgment, indicating that the false edge of the protective film tear was determined to cross or align with the process feature point coordinates in step S4, and the verification is passed; the second category is a false edge non-maintainment judgment, indicating that the false edge of the protective film tear was determined to not cross and not align with the process feature point coordinates in step S4, and the verification is failed, and it has been restored to a pending feature. The false edges of the protective film tear corresponding to the first category of verification results are collected to form a set of false edges to be removed. After the candidate false features are determined in step S2, the candidate false features are stored in the form of a set, and the set of candidate false features is used as the set of candidate false features to be removed. The multiple boundary features extracted in step S1 constitute the original boundary feature set. Each boundary feature in the original boundary feature set has a unique corresponding boundary feature identifier, which is assigned when extracted in step S1. The boundary feature identifier is in the form of a numerical sequence number or a string, and the boundary feature identifier remains unchanged during the processing of this batch of aluminum single panels.

[0067] The first removal operation removes all protective membrane tear pseudo-edges contained in the set of pseudo-edges to be removed from the original boundary feature set. Specifically, the removal operation iterates through each boundary feature in the original boundary feature set, checking if the feature appears in the set of pseudo-edges to be removed. If the feature does appear, it is removed from the original boundary feature set. If it does not appear, it is retained in the original boundary feature set. After the first removal operation, the original boundary feature set is updated to the boundary feature set after the first removal.

[0068] A second elimination operation is performed, removing all candidate pseudo-features from the boundary feature set after the first elimination. Specifically, each boundary feature in the boundary feature set after the first elimination is iterated over, and each boundary feature is checked to see if it appears in the candidate pseudo-feature set. If it does, it is removed from the boundary feature set after the first elimination. If it does not appear, it is retained in the boundary feature set. After the second elimination operation, the boundary feature set after the first elimination is updated to the boundary feature set after the second elimination.

[0069] The execution order of the first and second elimination operations can be interchanged. That is, candidate pseudo-features can be eliminated from the original boundary feature set first, and then the protective membrane tear pseudo-edges that maintain the pseudo-edge determination can be eliminated from the boundary feature set after elimination.

[0070] When there is an intersection between the set of pseudo-edges to be removed and the set of candidate pseudo-features to be removed, that is, when a certain boundary feature is determined to be a candidate pseudo-feature in step S2 and identified as a protective film tear pseudo-edge in step S3 and the pseudo-edge determination is maintained in step S4, the boundary feature is removed once in the first removal operation and the second removal operation. The final result will not be affected by repeated removal because when the boundary feature has been removed in the first removal operation, the boundary feature is no longer in the set of boundary features to be processed when the second removal operation is performed, and it will not be encountered again during traversal.

[0071] After the first and second rejection operations, the remaining boundary features in the boundary feature set after the second rejection constitute a reliable boundary feature set. Each boundary feature in the reliable boundary feature set satisfies the following conditions: the boundary feature was not identified as a candidate false feature in step S2, that is, the boundary feature successfully matched the sheet metal contour or predetermined cutting contour defined in the original tool path data in step S2; and the boundary feature was not identified and remained a protective film tear false edge in the combined processing of steps S3 and S4, that is, the boundary feature was either not identified as a protective film tear false edge in step S3, or it was identified as a protective film tear false edge in step S3 but failed the review in step S4 and was restored to a pending feature and was not included in the set of false edges to be rejected. The remaining boundary features in the reliable boundary feature set are indexed by the boundary feature identifier, and the pixel coordinate sequence of each boundary feature stored in step S1, as well as the row coordinates, column coordinates and original brightness values ​​of each pixel, are retained for use in step S6 when calculating the spatial transformation relationship.

[0072] In step S6, the boundary features contained in the reliable boundary feature set are used as spatial reference points. The rigid body transformation matrix that minimizes the positional deviation between the reliable boundary feature set and the corresponding contour in the original toolpath data is calculated. The rigid body transformation matrix is ​​used as a spatial transformation relationship. The spatial transformation relationship is applied to the coordinates of all toolpath points in the original toolpath data to generate the corrected machining path.

[0073] After constructing the credible boundary feature set in step S5, each boundary feature in the credible boundary feature set is stored in the form of a pixel coordinate sequence, which includes the row coordinates and column coordinates of each pixel. The pixel coordinate sequence of each boundary feature in the credible boundary feature set is transformed to the workpiece coordinate system. The transformation method is to use the pixel spacing calibration value of the imaging device to multiply the row coordinates by the pixel spacing calibration value to obtain the X-axis coordinate value in the workpiece coordinate system, and to multiply the column coordinates by the pixel spacing calibration value to obtain the Y-axis coordinate value in the workpiece coordinate system. The pixel spacing calibration value of the imaging device is pre-calculated by imaging a standard calibration plate and measuring the number of pixels corresponding to known dimensional features on the plate. The standard calibration plate surface has a geometric pattern with known physical dimensions. After the imaging device captures an image of the standard calibration plate, the number of pixels occupied by the geometric pattern in the positioning image is measured. The pixel spacing calibration value is obtained by dividing the known physical dimensions by the number of pixels. For example, if the geometric pattern on the standard calibration plate is a 20mm × 20mm square, and the number of pixels for the side length of this square in the positioning image is 200, then the pixel spacing calibration value is 0.10mm. Each boundary feature in the reliable boundary feature set corresponds to a sequence of coordinate points in the workpiece coordinate system.

[0074] The original toolpath data was retrieved in step S2. This data includes the sheet metal contour definition and the predetermined cutting contour definition. The contour point coordinate sequence of the sheet metal contour definition is represented in the workpiece coordinate system, as is the contour point coordinate sequence of the predetermined cutting contour definition. For each boundary feature in the reliable boundary feature set, the corresponding contour is searched in both the sheet metal contour definition and the predetermined cutting contour definition. The method for finding the corresponding contour is as follows: the coordinate point sequence of the boundary feature in the reliable boundary feature set in the workpiece coordinate system is shape-matched with the contour point coordinate sequence of the sheet metal contour definition; simultaneously, the coordinate point sequence of the boundary feature in the workpiece coordinate system is shape-matched with the contour point coordinate sequence of the predetermined cutting contour definition. The contour with the highest shape similarity to the boundary feature and exceeding a preset matching confirmation threshold is determined as the corresponding contour of the boundary feature in the original toolpath data. The preset matching confirmation threshold is set by using the value of the preset matching threshold from step S2, or by taking a preset multiple of the preset matching threshold, such as multiplying the preset matching threshold by a value between 0.8 and 1.2.

[0075] After determining the corresponding contour in the original toolpath data for each boundary feature in the reliable boundary feature set, the rigid body transformation matrix is ​​calculated using all boundary features in the reliable boundary feature set as spatial reference points. The rigid body transformation matrix includes rotation and translation components. The calculation process involves iteratively solving for rotation and translation parameters that minimize the sum of positional deviations between all spatial reference points and their corresponding contours. Positional deviation is defined as the Euclidean distance between the coordinates of a spatial reference point after transformation by the rigid body transformation matrix and the coordinates of the nearest point on the corresponding contour. The sum of positional deviations is either the sum of the squares of the positional deviations of all spatial reference points, or the sum of the absolute values ​​of the positional deviations of all spatial reference points. The initial values ​​for the iterative solution are set as follows: the initial value of the rotation parameter is set to zero rotation angle, i.e., the initial value of the rotation parameter is 0 degrees or 0 radians; the initial value of the translation parameter is set to the offset vector between the geometric centers of all boundary features in the reliable boundary feature set and the geometric centers of all corresponding contours. The geometric centers of all boundary features in the reliable boundary feature set are two-dimensional coordinate points obtained by averaging the X-axis coordinates and Y-axis coordinates of all coordinate points in the coordinate point sequence of all boundary features in the workpiece coordinate system. The geometric centers of all corresponding contours are two-dimensional coordinate points obtained by averaging the X-axis coordinates and Y-axis coordinates of all coordinate points in the contour point coordinate sequence of all corresponding contours. The iteration termination condition is that the decrease in the sum of position deviations between two adjacent iterations is lower than the preset iteration convergence threshold. The preset iteration convergence threshold is set by taking a preset proportion of the pixel spacing calibration value, for example, taking the pixel spacing calibration value multiplied by a value between 0.01 and 0.10. The iteration terminates when the decrease in the sum of position deviations after a certain iteration is lower than the preset iteration convergence threshold.

[0076] After iterative solution, the rotation and translation parameters that minimize the total positional deviation are obtained. These parameters are then combined to form a rigid body transformation matrix. The rigid body transformation matrix is ​​a 3x3 matrix. The top 2x2 elements are the rotation matrix, where the first element in the first row and first column is the cosine of the rotation angle, the second element in the first row and second column is the negative of the sine of the rotation angle, the first element in the second row and first column is the sine of the rotation angle, and the second element in the second row and second column is the cosine of the rotation angle. The top 2x1 elements are the translation vector, where the first element in the first row is the X-axis translation, and the second element in the second row is the Y-axis translation. The bottom 1x2 elements are zero, and the bottom 1x1 element is 1. The rigid body transformation matrix is ​​used as a spatial transformation relation.

[0077] The spatial transformation relation is applied to the coordinates of all toolpath points in the original toolpath data to generate the corrected machining path. The original toolpath data contains multiple toolpath trajectories, each composed of a sequence of toolpath point coordinates, represented by the X-axis and Y-axis coordinates in the workpiece coordinate system. For each toolpath point coordinate in each toolpath trajectories, the rotation matrix in the spatial transformation relation is multiplied by the toolpath point coordinate, and then the translation vector in the spatial transformation relation is added to the product to obtain the transformed toolpath point coordinates. The specific operation of multiplying the rotation matrix with the toolpath point coordinates is as follows: multiply the X-axis coordinate of the toolpath point by the element in the first row and first column of the rotation matrix, and add the Y-axis coordinate of the toolpath point by the element in the first row and second column of the rotation matrix to obtain the transformed X-axis component. Then, multiply the X-axis coordinate of the toolpath point by the element in the second row and first column of the rotation matrix, and add the Y-axis coordinate of the toolpath point by the element in the second row and second column of the rotation matrix to obtain the transformed Y-axis component. Next, add the X-axis component of the transformed coordinates to the element in the first row of the translation vector to obtain the transformed X-axis coordinate value. Finally, add the Y-axis component of the transformed coordinates to the element in the second row of the translation vector to obtain the transformed Y-axis coordinate value. This transformation operation is performed sequentially on all toolpath point coordinates for all toolpath trajectories. All transformed toolpath point coordinates are reconnected according to the trajectory segment order and point order within the trajectory segments in the original toolpath data to form the corrected machining path. The corrected machining path spatially matches the actual contour of the aluminum panel to be processed.

[0078] Example 2: Figure 2 A schematic diagram of the automatic path correction system for aluminum single-panel engraving based on positioning image feedback of the present invention is given. The automatic path correction system for aluminum single-panel engraving based on positioning image feedback includes the following modules:

[0079] The image acquisition module is used to acquire positioning images containing the state of the protective film on the surface of aluminum panels and extract multiple boundary features;

[0080] The feature matching module is used to retrieve the sheet metal contour and the predetermined cutting contour from the original tool path data, match each boundary feature with the sheet metal contour and the predetermined cutting contour, and determine the boundary features that cannot be matched with the corresponding contour as candidate pseudo features.

[0081] The brightness analysis module is used to extract the brightness distribution along the normal direction of each boundary feature and calculate the brightness gradient asymmetry. The brightness gradient asymmetry is lower than the preset symmetry threshold and is identified as a protective film tear pseudo edge superimposed on the real boundary.

[0082] The geometric verification module is used to retrieve the coordinates of process feature points in the original toolpath data and verify the false edges of the protective film tear. If its geometric extension path crosses or aligns with the coordinates of the process feature points, the false edge judgment is maintained.

[0083] The feature filtering module is used to remove protective membrane tear pseudo-edges and candidate pseudo-features from multiple boundary features after verification to maintain the pseudo-edge determination, and to form a credible boundary feature set from the remaining boundary features.

[0084] The path generation module is used to calculate the spatial transformation relationship based solely on the set of credible boundary features, and to perform an overall transformation on the original toolpath data based on the spatial transformation relationship to generate the corrected machining path.

[0085] All calculations involved in the embodiments are dimensionless numerical calculations, and the preset parameters and thresholds in the calculations are set by those skilled in the art according to the actual situation.

[0086] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0087] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and inventive constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0088] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0089] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or modules may be electrical, mechanical, or other forms.

[0090] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0091] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An automatic correction method for engraving paths on aluminum single-panel panels based on positioning image feedback, characterized in that, Includes the following steps: S1: Obtain a positioning image containing the state of the protective film on the surface of the aluminum panel and extract multiple boundary features; S2: Retrieve the sheet metal contour and the predetermined cutting contour from the original toolpath data, match each boundary feature with the sheet metal contour and the predetermined cutting contour, and determine the boundary features that cannot be matched with the corresponding contour as candidate pseudo features. S3: Extract the brightness distribution along the normal direction for each boundary feature and calculate the brightness gradient asymmetry. Identify those with brightness gradient asymmetry below the preset symmetry threshold as protective film tear pseudo-edges superimposed on the real boundary. S4: Retrieve the coordinates of process feature points in the original toolpath data and verify the false edges of the protective film tear. If the geometric extension path crosses or aligns with the coordinates of the process feature points, the false edge judgment is maintained. S5: Remove the protective membrane tear pseudo-edge and candidate pseudo-features that maintain the pseudo-edge determination after verification from multiple boundary features, and form a credible boundary feature set from the remaining boundary features. S6: Calculate the spatial transformation relationship using only the credible boundary feature set as a constraint, and perform an overall transformation on the original toolpath data based on the spatial transformation relationship to generate the corrected machining path.

2. The automatic correction method for aluminum single-panel engraving path based on positioning image feedback according to claim 1, characterized in that, Acquire a positioning image containing the state of the protective film on the surface of the aluminum single panel, and extract multiple boundary features, including: acquiring a positioning image containing the state of the protective film on the surface of the aluminum single panel, identifying pixels with abrupt changes in brightness in the positioning image, connecting the pixels with abrupt changes in brightness to form multiple continuous lines, and extracting the continuous lines that are closed or cross the boundary of the positioning image as boundary features.

3. The automatic correction method for aluminum single-panel engraving path based on positioning image feedback according to claim 1, characterized in that, The process involves retrieving the sheet metal contour and the predetermined cutting contour from the original toolpath data, matching each boundary feature with the sheet metal contour and the predetermined cutting contour, and identifying boundary features that cannot be matched with the corresponding contour as candidate pseudo features. This includes: retrieving the sheet metal contour and the predetermined cutting contour defined in the original toolpath data, converting the sheet metal contour into a first polygon representation, converting the predetermined cutting contour into a second polygon representation, converting each boundary feature into a third polygon representation, calculating the shape similarity between the third polygon representation and the first polygon representation, and the shape similarity between the third polygon representation and the second polygon representation, and identifying boundary features whose shape similarity is lower than a preset matching threshold as candidate pseudo features.

4. The automatic correction method for aluminum single-panel engraving path based on positioning image feedback according to claim 1, characterized in that, The brightness distribution of each boundary feature is extracted along the normal direction and the brightness gradient asymmetry is calculated. The brightness gradient asymmetry below a preset symmetry threshold is identified as a protective film tear pseudo-edge superimposed on the real boundary. This includes: taking the determined candidate pseudo-features as the processing object, taking multiple sampling points along the extension direction of the candidate pseudo-features at a preset step size, extracting the brightness value sequence along the normal direction of the candidate pseudo-features at each sampling point, calculating the ratio of the gradient magnitude of the brightness falling edge to the gradient magnitude of the brightness rising edge in the brightness value sequence, taking the ratio as the brightness gradient asymmetry, and identifying the candidate pseudo-features with brightness gradient asymmetry below the preset symmetry threshold as a protective film tear pseudo-edge superimposed on the real boundary.

5. The automatic correction method for aluminum single-panel engraving path based on positioning image feedback according to claim 4, characterized in that, The calculation of the ratio of the gradient magnitude of the falling edge to the gradient magnitude of the rising edge in the brightness value sequence includes: at each sampling point of the candidate pseudo-feature, traversing the brightness value sequence from one side of the candidate pseudo-feature to the other along the normal direction, identifying the continuous pixel segments where the brightness value transitions from the high brightness range to the low brightness range as the falling edge and recording their gradient magnitude, identifying the continuous pixel segments where the brightness value transitions from the low brightness range to the high brightness range as the rising edge and recording their gradient magnitude, and dividing the gradient magnitude of the falling edge by the gradient magnitude of the rising edge to obtain the brightness gradient asymmetry.

6. The automatic correction method for aluminum single-panel engraving path based on positioning image feedback according to claim 1, characterized in that, The coordinates of process feature points in the original toolpath data are retrieved to verify the pseudo-edge of the protective film tear. If its geometric extension path crosses or aligns with the coordinates of the process feature points, the pseudo-edge determination is maintained. This includes: retrieving the coordinates of the chamfer vertex and the contour tip recorded in the original toolpath data as the coordinates of the process feature points; generating a geometric extension path along the extension direction of the pseudo-edge of the protective film tear; calculating the minimum distance between the geometric extension path and the coordinates of the process feature points; and determining the pseudo-edge of the protective film tear that the minimum distance is lower than a preset distance threshold as crossing or aligning with the coordinates of the process feature points, thus maintaining the pseudo-edge determination for the pseudo-edge of the protective film tear.

7. The automatic correction method for aluminum single-panel engraving path based on positioning image feedback according to claim 6, characterized in that, Generating a geometric extension path along the extension direction of the protective film tear pseudo edge includes: obtaining the two endpoints of the protective film tear pseudo edge, extending outward from the two endpoints along the tangent direction of the protective film tear pseudo edge at the endpoints respectively, with the extension length being a preset proportion of the total length of the protective film tear pseudo edge, and connecting the protective film tear pseudo edge and its two end extension segments to form a geometric extension path.

8. The automatic correction method for aluminum single-panel engraving path based on positioning image feedback according to claim 1, characterized in that, The protective membrane tear pseudo-edges and candidate pseudo-features that maintain the pseudo-edge determination after verification are removed from multiple boundary features. The remaining boundary features constitute a credible boundary feature set, including: removing the protective membrane tear pseudo-edges that maintain the pseudo-edge determination from multiple boundary features, removing the boundary features that are determined to be candidate pseudo-features from multiple boundary features, and taking all the remaining boundary features after removal as the credible boundary feature set.

9. The automatic correction method for engraving path of aluminum single-panel based on positioning image feedback according to claim 1, characterized in that, The spatial transformation relationship is calculated using only the credible boundary feature set as a constraint. Based on the spatial transformation relationship, the original toolpath data is transformed as a whole to generate the corrected machining path. This includes: using the boundary features contained in the credible boundary feature set as spatial reference points, calculating the rigid body transformation matrix that minimizes the positional deviation between the credible boundary feature set and the corresponding contours in the original toolpath data, using the rigid body transformation matrix as the spatial transformation relationship, applying the spatial transformation relationship to the coordinates of all toolpath points in the original toolpath data, and generating the corrected machining path.

10. An automatic path correction system for aluminum single-panel engraving based on positioning image feedback, used to implement the automatic path correction method for aluminum single-panel engraving based on positioning image feedback as described in any one of claims 1-9, characterized in that, Includes the following modules: The image acquisition module is used to acquire positioning images containing the state of the protective film on the surface of aluminum panels and extract multiple boundary features; The feature matching module is used to retrieve the sheet metal contour and the predetermined cutting contour from the original tool path data, match each boundary feature with the sheet metal contour and the predetermined cutting contour, and determine the boundary features that cannot be matched with the corresponding contour as candidate pseudo features. The brightness analysis module is used to extract the brightness distribution along the normal direction of each boundary feature and calculate the brightness gradient asymmetry. The brightness gradient asymmetry is lower than the preset symmetry threshold and is identified as a protective film tear pseudo edge superimposed on the real boundary. The geometric verification module is used to retrieve the coordinates of process feature points in the original toolpath data and verify the false edges of the protective film tear. If its geometric extension path crosses or aligns with the coordinates of the process feature points, the false edge judgment is maintained. The feature filtering module is used to remove protective membrane tear pseudo-edges and candidate pseudo-features from multiple boundary features after verification to maintain the pseudo-edge determination, and to form a credible boundary feature set from the remaining boundary features. The path generation module is used to calculate the spatial transformation relationship based solely on the set of credible boundary features, and to perform an overall transformation on the original toolpath data based on the spatial transformation relationship to generate the corrected machining path.

Images