Embroidery control method and device based on vision system

Abstract

The application provides an embroidery control method and device based on a vision system, and relates to the technical field of embroidery processing. The method comprises the following steps: acquiring an image of an object to be embroidered through a vision system; identifying actual shape parameters of the object to be embroidered based on the image; adaptively reconstructing embroidery pattern data based on the actual shape parameters, so that the reconstructed pattern data matches the actual shape of the object to be embroidered; and controlling an embroidery execution unit to perform an embroidery operation according to the reconstructed embroidery pattern data. The application uses vision recognition to acquire actual shape parameters of a fabric and dynamically reconstructs pattern data, realizes adaptive changes of the pattern data according to the fabric state, effectively solves the embroidery adaptation problem caused by differences in the fabric contour or size deformation, and significantly improves the embroidery precision and production efficiency.

Description

Technical Field

[0001] This invention relates to the field of embroidery processing technology, and specifically to an embroidery control method and device based on a vision system. Background Technology

[0002] Currently, in the field of embroidery machine processing, traditional embroidery operations typically rely on manual positioning or mechanical template matching. When faced with complex scenarios such as patch embroidery, feature point positioning, and fabric deformation, existing technologies often have significant limitations. For example, in patch embroidery, workers need to manually align the edges of the patches and draw the embroidery pattern, which is not only inefficient but also prone to human error leading to scrapped embroidery. For fabrics with feature points, the accuracy of manual positioning for embroidery is difficult to control, and it cannot adapt to the dimensional deformation caused by fabric stretching and shrinking, often resulting in pattern misalignment and poor fabric fit. As the embroidery industry's requirements for precision and efficiency continue to increase, the problems of traditional embroidery machines—relying on manual labor, low precision, and poor adaptability—are becoming increasingly prominent, making it difficult to meet the demands of large-scale, intelligent production. Summary of the Invention

[0003] To address the problems of low matching degree between embroidery patterns and actual fabric shape in existing technologies, and poor accuracy and low efficiency due to reliance on manual positioning, this application proposes an embroidery control method and device based on a vision system. This method utilizes visual recognition to obtain actual fabric shape parameters and dynamically reconstructs pattern data, thereby achieving automation and intelligent adaptation of the embroidery process.

[0004] To achieve the above objectives, the present invention adopts the following technical solution: An embroidery control method based on a vision system includes: acquiring an image of an object to be embroidered through a vision system; identifying actual shape parameters of the object to be embroidered based on the image; adaptively reconstructing embroidery pattern data based on the actual shape parameters so that the reconstructed pattern data matches the actual shape of the object to be embroidered; and controlling an embroidery execution unit to perform embroidery work according to the reconstructed embroidery pattern data.

[0005] The above solution obtains the actual shape parameters of the object to be embroidered through visual recognition, and adaptively reconstructs the embroidery pattern data accordingly. This breaks the traditional mode of fixed pattern data and mechanical position adjustment, and realizes the adaptive change of pattern data with the fabric state. It effectively solves the embroidery adaptation problem caused by differences in fabric contour or size deformation, and significantly improves embroidery accuracy and production efficiency.

[0006] Preferably, the actual morphological parameters include at least one of the outline of the object to be embroidered, the position of feature points, or the amount of deformation.

[0007] Preferably, the adaptive reconstruction includes at least one of coordinate transformation, size scaling, or stitch density adjustment of the embroidery pattern data.

[0008] Preferably, the stitch density adjustment includes: adjusting the stitch step length based on the deformation amount in the actual shape parameters so that the stitch density matches the degree of deformation of the object to be embroidered.

[0009] Preferably, the deformation amount is obtained by: obtaining a standard template image corresponding to the object to be embroidered; comparing the image of the object to be embroidered with the standard template image at the pixel level, and calculating the deformation amount.

[0010] Preferably, the actual shape parameter is the position of the feature point of the object to be embroidered; based on the position of the feature point, the actual embroidery position of the embroidery pattern data on the object to be embroidered is determined by coordinate conversion.

[0011] Preferably, the actual morphological parameter is the deformation amount of the object to be embroidered; based on the deformation amount, the size ratio and stitch density of the embroidery pattern data are adaptively reconstructed so that the reconstructed pattern data matches the deformation degree of the object to be embroidered.

[0012] Preferably, before acquiring the image of the object to be embroidered, the method further includes calibrating the image acquisition unit, wherein the calibration includes at least one of flatness calibration, distortion correction, and coordinate origin calibration.

[0013] Furthermore, the present invention also provides a vision-based embroidery control device, applicable to a vision-based embroidery control method as described above, comprising: an image acquisition unit configured to acquire an image of an object to be embroidered; an AI processing unit configured to identify actual morphological parameters of the object to be embroidered based on the image; adaptively reconstructing embroidery pattern data based on the actual morphological parameters, so that the reconstructed pattern data matches the actual morphology of the object to be embroidered; and an embroidery execution unit configured to perform embroidery work according to the reconstructed embroidery pattern data.

[0014] Preferably, the AI ​​processing unit includes a processor and a memory, the memory storing a computer program, and the processor executing the computer program to realize the functions of recognizing the actual shape parameters of the object to be embroidered and adaptively reconstructing the embroidery pattern data.

[0015] Beneficial effects: This invention uses a vision system to acquire images of the object to be embroidered and identify its actual shape parameters. It then adaptively reconstructs the embroidery pattern data, allowing the pattern data to adjust itself according to the fabric's contour, feature point position, or deformation. This control logic, which reconstructs data through visual perception, differs from traditional simple position alignment, achieving precise matching between pattern data and the actual fabric shape. Especially when the fabric deforms, the dual reconstruction of the pattern data's size proportions and stitch density effectively avoids problems such as pattern deformation or uneven density, significantly improving embroidery quality, reducing reliance on skilled workers, and increasing production efficiency and automation. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating the embroidery control method based on a vision system provided in an embodiment of the present invention; Figure 2 This is a comparative schematic diagram showing the embroidery pattern data before and after adaptive reconstruction provided in an embodiment of the present invention. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0018] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention.

[0019] Example 1: like Figure 1 As shown, this embodiment provides an embroidery control method based on a vision system. This method solves the problems of low matching degree and high scrap rate caused by fixed pattern data and variable fabric shapes in traditional embroidery processes through closed-loop control of visual perception and data reconstruction. Specifically, it includes the following steps: Step S1: Obtain an image of the object to be embroidered through a vision system.

[0020] Specifically, a vision system typically includes an image acquisition unit and a corresponding light source system. The image acquisition unit can be, for example, an industrial camera, a line scan camera, or an area scan camera. During this step, the image acquisition unit photographs the object to be embroidered, placed on the embroidery frame, acquiring raw image data containing its surface texture, contours, feature points, or overall shape. To ensure image quality, the light source system can provide uniform illumination, avoiding shadows or reflections that could interfere with subsequent recognition processes. It should be understood that the timing of image acquisition can be set according to actual process requirements, such as taking a global shot before embroidery begins, or taking localized real-time shots during the embroidery process.

[0021] Step S2: Based on the image, identify the actual shape parameters of the object to be embroidered.

[0022] Specifically, the actual morphological parameters are a set of key data characterizing the current physical state of the object to be embroidered. This step utilizes image processing algorithms to analyze the image acquired in step S1. For example, edge detection algorithms are used to extract the contour coordinates of the object to be embroidered; feature matching algorithms are used to identify the position coordinates of preset marker points or fabric texture feature points; or comparative analysis is used to calculate the deformation of the fabric relative to its standard state. In this embodiment, the identification process is not limited to a single parameter, but rather extracts necessary parameters from the image that can guide pattern adjustments based on the specific embroidery scene requirements. This breaks the limitations of traditional technologies that rely solely on manual visual inspection or mechanical positioning, providing precise data support for subsequent data reconstruction.

[0023] Step S3: Based on the actual shape parameters, adaptively reconstruct the embroidery pattern data so that the reconstructed pattern data matches the actual shape of the object to be embroidered.

[0024] Specifically, adaptive reconstruction is the core step of this embodiment, essentially a dynamic correction of the embroidery pattern data. Embroidery pattern data typically includes control information such as stitch coordinates, stitch type, and stitch length. In traditional techniques, pattern data are pre-set fixed values; once the fabric deforms or shifts position, the embroidered pattern will be distorted. In this embodiment, the control system calculates and modifies the pattern data in real time based on the actual shape parameters identified in step S2. This reconstruction can be a translation or rotation of the coordinate system in a positioning embroidery scenario, a scaling of the overall size in a scenario where the fabric shrinks, or a local adjustment of the stitch density in a scenario where the fabric is stretched locally. Through adaptive reconstruction, the digitized pattern data can match the physical fabric shape, thereby ensuring a perfect visual and structural fit of the final embroidery.

[0025] Step S4: Control the embroidery execution unit to perform embroidery work according to the reconstructed embroidery pattern data.

[0026] Specifically, the embroidery execution unit includes hardware modules such as an embroidery frame drive mechanism and a needle bar drive mechanism. The control system converts the reconstructed pattern data into motor control commands, driving the embroidery frame to move and the needle to actuate, completing the actual embroidery actions. Because the pattern data has been adaptively reconstructed, the embroidery execution unit can accurately adapt to the actual state of the fabric during operation, without the need for manual intervention, significantly improving the automation level of embroidery and the quality of the finished product.

[0027] This embodiment establishes a control logic for visual perception, data reconstruction, and precise execution through the above steps. This logic no longer treats the fabric as an idealized rigid plane, but rather as a flexible object with random shape changes. The visual system captures these changes in real time and provides feedback to correct the pattern data, thus solving the embroidery quality problems caused by fabric deformation or positioning deviations at the source.

[0028] Example 2: This embodiment further explains the specific implementation of adaptive reconstruction based on Embodiment 1. The adaptive reconstruction includes at least one of coordinate transformation, size scaling, or stitch density adjustment of the embroidery pattern data.

[0029] Specifically, coordinate transformation is typically used in positioning embroidery, where the shape and size of the pattern data itself remain unchanged, and its position in the coordinate system is changed only by translation or rotation. Size scaling, on the other hand, is used in scenarios where the fabric undergoes uniform deformation, i.e., the geometric outline of the pattern is enlarged or reduced proportionally. However, in actual embroidery processes, simply scaling the geometric dimensions is often insufficient to solve the quality problems caused by changes in the physical form of the fabric. For example, when the fabric is stretched, if the pattern is simply enlarged proportionally without adjusting the stitch length, the stitch density per unit area will significantly decrease, resulting in the fabric background showing through. Conversely, when the fabric shrinks, if the pattern is simply reduced without adjusting the stitch length, the stitch density per unit area will be too high, causing the fabric to become stiff or even wrinkled.

[0030] Therefore, this embodiment focuses on the technical means of adjusting stitch density. The stitch density adjustment includes: adjusting the stitch step length based on the deformation amount in the actual shape parameters, so that the stitch density matches the degree of deformation of the object to be embroidered.

[0031] The stitch step length refers to the distance the needle bar of the embroidery machine moves between two adjacent needle points. The larger the step length, the sparser the stitches per unit length; the smaller the step length, the denser the stitches. This embodiment achieves dynamic correction of the physical properties of embroidery by establishing a correlation between deformation and stitch step length.

[0032] For example, when the vision system detects that the object to be embroidered is stretched, the control system calculates the stretch ratio and increases the stitch step length accordingly. This means that within the same geometric distance, the number of stitches performed by the embroidery machine is reduced, thereby lowering the stitch density and compensating for areas thinned by fabric stretching. This effectively prevents the fabric from bunching or wrinkling due to overly dense stitches. Conversely, when shrinkage deformation of the fabric is detected, the control system reduces the stitch step length and increases the number of stitches per unit length, increasing the stitch density. This avoids insufficient coverage or exposed fabric due to fabric shrinkage.

[0033] It should be understood that this stitch density adjustment based on physical deformation differs from the simple image scaling processing in existing technologies. Image scaling only changes the visual size of the pattern and cannot alter the physical needle-feeding logic of the embroidery machine. This embodiment, by modifying the underlying embroidery control data, resolves the mismatch between fabric deformation and embroidery density at the physical level, significantly improving the texture and pass rate of the embroidered products.

[0034] Example 3: This embodiment, based on Embodiment 1, further details the high-precision data acquisition and processing process of the vision system. Before the embroidery operation begins, in order to eliminate the influence of hardware installation errors on visual recognition accuracy, the system must be initialized and calibrated. Before acquiring the image of the object to be embroidered, the method further includes calibrating the image acquisition unit, wherein the calibration includes at least one of flatness calibration, distortion correction, and coordinate origin calibration.

[0035] Specifically, during the installation of the image acquisition unit, there may be a slight angular deviation between its optical axis and the embroidery frame plane, or the lens itself may have optical distortion. These factors can cause geometric distortion between the acquired image and the actual object. Without calibration, the subsequently calculated deformation will include systematic errors, leading to inaccurate pattern reconstruction. Flatness calibration corrects perspective errors caused by the camera's optical axis not being perpendicular to the embroidery frame's working plane, ensuring consistent magnification across the image. Distortion correction eliminates barrel or pincushion distortion at the lens edges, restoring the image's true geometry. Coordinate origin calibration establishes a mapping between the image coordinate system and the embroidery machine's mechanical coordinate system, ensuring that the visually recognized position is accurately translated into the needle's movement trajectory. These calibration steps provide a data benchmark for subsequent high-precision visual adaptation, guaranteeing the accuracy of embroidery control from the outset.

[0036] After system calibration is completed, this embodiment provides a specific calculation method for obtaining fabric deformation. The deformation is obtained by: acquiring a standard template image corresponding to the object to be embroidered; comparing the image of the object to be embroidered with the standard template image at the pixel level, and calculating the deformation.

[0037] The standard template image refers to a reference image captured under the standard condition of the fabric, or a design drawing image pre-stored in the system. In actual operation, the vision system first captures a real-time image of the fabric to be embroidered on the current embroidery frame, and then calls the pre-stored standard template image. Using an image registration algorithm, the two images are aligned at the pixel level, and then their geometric feature differences are compared pixel-by-pixel or region-by-region. For example, if the fabric is stretched in the X-axis direction, the pixel spacing of the feature points in the real-time image in the X-axis direction will be greater than the corresponding spacing in the standard template image; if the fabric undergoes local irregular deformation, the grid lines or feature point sets in the image will exhibit non-linear offsets. By calculating the statistical values ​​or distribution fields of these offsets, the system can accurately determine the amount of fabric deformation. This pixel-level comparison-based calculation method can capture micron-level deformation details and has extremely high sensitivity to uneven fabric deformation, providing a precise quantitative basis for the stitch density adjustment described in Example 2, thus achieving a technological leap from qualitative identification to quantitative calculation.

[0038] Example 4: This embodiment further explains the positioning embroidery scenario based on Embodiment 1. In this scenario, the actual morphological parameters are the feature point positions of the object to be embroidered.

[0039] Specifically, feature point locations refer to reference coordinate points that characterize the actual placement of the object to be embroidered within the embroidery frame. These feature points can be special markings pre-printed on the fabric by the operator, such as cross marks or dot marks; or they can be inherent natural features of the fabric itself, such as the intersections of fabric textures, the intersections of seams, buttonholes in clothing, or corner points of specific patterns. The vision system uses image recognition algorithms to accurately extract the positions of these feature points in the image coordinate system, and then maps them to the mechanical coordinate system of the embroidery machine through a coordinate transformation matrix.

[0040] Based on the location of the feature points, the actual embroidery position of the embroidery pattern data on the object to be embroidered is determined by coordinate conversion.

[0041] Coordinate transformation refers to establishing a mapping relationship between the image coordinate system and the embroidery machine's mechanical coordinate system. Because the placement of fabric on the embroidery frame often involves random translation or rotational deviations, the preset embroidery starting point and the actual embroidery area on the fabric often cannot be aligned. In this embodiment, the AI ​​processing unit first identifies at least two non-collinear feature points and calculates the coordinate transformation parameters based on the deviation between the actual coordinates of the feature points and the preset coordinates. Subsequently, the system performs corresponding translation and rotation on the overall coordinates of the embroidery pattern data to determine the actual needle placement position of the pattern on the fabric.

[0042] It is particularly important to emphasize that the adaptive reconstruction described in this embodiment mainly involves a transformation of the coordinate system. That is, the shape, size, and relative positional relationships of the internal stitches of the pattern data itself remain unchanged; only the absolute position and angle of the pattern on the fabric are altered. This is fundamentally different from the stitch density adjustment involved in Embodiment 2 and the size ratio reconstruction involved in subsequent embodiments. Positioning embroidery is typically suitable for applications where the fabric itself is stable and has not undergone stretching or shrinkage deformation, but a specific pattern needs to be precisely embroidered in a specific location on the fabric. For example, in garment embroidery, a brand logo needs to be precisely embroidered in a specific position on a pocket flap. In this case, only coordinate conversion is needed through feature point positioning, without scaling or density adjustment of the logo's pattern data. In this way, the system can automatically correct positional errors caused by manual fabric placement, significantly improving the automation level and finished product qualification rate of positioning embroidery operations.

[0043] Example 5: This embodiment further explains the AI ​​vision adaptation scenario based on Embodiment 1. In this scenario, the actual shape parameter is the deformation amount of the object to be embroidered.

[0044] Specifically, this scenario primarily addresses situations where fabrics undergo non-uniform deformation during processing, storage, or transportation. Unlike the positioning embroidery scenario described in Example 4, which assumes the fabric's shape remains unchanged only in position, the AI ​​visual adaptation scenario deals with situations where the fabric's physical dimensions have been stretched, contracted, or twisted. In this case, simply adjusting the pattern's coordinates as in Example 4 will result in severe distortion due to fabric deformation; for example, a circular pattern might become an ellipse, or a square pattern a rhombus. Therefore, a deeper reconstruction of the pattern data is necessary.

[0045] Based on the deformation amount, the size ratio and stitch density of the embroidery pattern data are adaptively reconstructed so that the reconstructed pattern data matches the degree of deformation of the object to be embroidered.

[0046] The reconstruction of size proportions refers to scaling and correcting the overall outline of the pattern data according to the stretching or shrinking ratio of the fabric. For example, if the vision system identifies that the fabric is stretched in the warp and contracted in the weft, the system will enlarge the warp dimension of the pattern data accordingly and shrink the weft dimension accordingly to compensate for the fabric deformation. However, simply scaling the geometric dimensions is not enough. As described in Example 2, fabric deformation is inevitably accompanied by changes in density. If the fabric is stretched, the yarn density per unit area decreases. If embroidery is still done at the original density, the embroidery thread will not be able to completely cover the fabric, resulting in exposed fabric. Conversely, if the fabric shrinks, the density increases, and embroidery at the original density will cause the fabric to wrinkle. Therefore, in this embodiment, while reconstructing the size proportions, the stitch density must be reconstructed simultaneously. The system adjusts the stitch step length synchronously according to the amount of deformation, increasing the step length in the stretched area to reduce the stitch density and decreasing the step length in the contracted area to increase the density, thereby achieving a dual match between geometric shape and physical texture.

[0047] Combination Figure 2 As shown, Figure 2 This diagram illustrates the comparison before and after adaptive reconstruction of embroidery pattern data. Label 1 represents the standard fabric shape, and label 2 represents the original pattern. When the new fabric shape (label 3) is used in actual operation, the vision system recognizes a significant deformation compared to the standard fabric. At this point, the system automatically generates the visually AI-processed pattern (label 4). It is evident that the processed pattern maintains the same outline size as the new fabric shape, and its internal stitch density is adaptively adjusted according to the degree of fabric deformation. This reconstruction is not a simple image scaling but a data regeneration based on the characteristics of the embroidery process, ensuring that the final embroidered product still presents a perfect visual effect and physical feel on deformed fabric. This embodiment, through the above dual reconstruction logic, solves the industry pain point of traditional embroidery machines' inability to adapt to fabric deformation, significantly improving the pass rate and adaptability of embroidered products.

[0048] Example 6: This embodiment provides an embroidery control device based on a vision system. This device is applicable to the embroidery control method described in the foregoing embodiments of this invention, achieving intelligent control of embroidery operations through the collaborative work of hardware architecture and software algorithms. The device mainly comprises three core modules: an image acquisition unit, an AI processing unit, and an embroidery execution unit.

[0049] The image acquisition unit is configured to acquire an image of the object to be embroidered.

[0050] Specifically, the image acquisition unit is responsible for acquiring raw visual data. In this embodiment, the image acquisition unit is preferably an industrial camera, mounted on the gantry of the embroidery machine, with its lens aimed at the embroidery frame area. To ensure the clarity and contrast of the acquired images, the industrial camera is typically equipped with a light source system to eliminate ambient light interference and shadows. It should be understood that the industrial camera can be an area scan camera, used to capture panoramic images of the static fabric; or it can be a line scan camera, used in conjunction with the fabric conveying mechanism for dynamic scanning and acquisition. The image acquisition unit communicates with the AI ​​processing unit through a high-speed data transmission interface, transmitting the acquired raw image data to the latter for processing in real time. High-precision image acquisition provides a high-quality data source for subsequent morphological parameter recognition.

[0051] The AI ​​processing unit is configured to identify the actual shape parameters of the object to be embroidered based on the image; and to adaptively reconstruct the embroidery pattern data based on the actual shape parameters so that the reconstructed pattern data matches the actual shape of the object to be embroidered.

[0052] Specifically, the AI ​​processing unit is responsible for core data calculation and logic control. In this embodiment, the AI ​​processing unit is manifested as a control system, which is typically integrated into the electrical control box of the embroidery machine. The control system receives image data from the image acquisition unit and executes preset image processing algorithms and logical operations. For example, the control system calls a pre-stored visual recognition algorithm model in the memory to perform edge detection, feature point matching, or pixel-level comparison on the image, thereby identifying the actual morphological parameters of the fabric, such as its outline, feature point positions, or deformation. Subsequently, based on the recognition results, the control system performs adaptive reconstruction calculations on the pre-stored embroidery pattern data, including operations such as coordinate transformation, size scaling, or stitch step length adjustment, to generate new pattern data adapted to the current fabric state.

[0053] The embroidery execution unit is configured to perform embroidery operations based on the reconstructed embroidery pattern data.

[0054] Specifically, the embroidery execution unit is responsible for converting digitized pattern data into physical embroidery. The embroidery execution unit mainly includes mechanical components such as an embroidery frame, a needle bar drive mechanism, a rotary hook mechanism, and a main spindle motor. The embroidery frame is used to tension and fix the fabric to be embroidered, and moves in a plane under the drive of X and Y axis stepper motors; the needle bar drive mechanism controls the up and down embroidery needle movement. The control system converts the reconstructed pattern data into motor control pulse signals, and precisely controls the movement trajectory of the embroidery frame and the needle placement position through a driver, thereby creating a pattern on the fabric that perfectly matches its actual shape. Through the collaborative operation of the above hardware architecture, this embodiment achieves a deep integration of visual recognition and embroidery execution, solving the problem that traditional embroidery machines cannot adapt to fabric deformation.

[0055] Furthermore, the AI ​​processing unit includes a processor and a memory, the memory storing a computer program, and the processor executing the computer program to realize the functions of recognizing the actual shape parameters of the object to be embroidered and adaptively reconstructing the embroidery pattern data.

[0056] Specifically, the processor can be a general-purpose microprocessor (such as a CPU), a digital signal processor (DSP), or an application-specific integrated circuit (ASIC)—a chip with data processing capabilities. The memory includes random access memory (RAM) and read-only memory (ROM), used to store the operating system, visual algorithm library, pattern database, and the aforementioned computer programs. When the device is running, the processor reads and executes the computer program instructions in the memory, performing a series of complex operations such as image processing, parameter calculation, and pattern reconstruction according to the logic flow described in the preceding embodiments. This hardware-software integrated architecture design ensures both high operational efficiency and stability, while also possessing good scalability, facilitating subsequent software upgrades to optimize algorithms or add new functions.

[0057] Example 7: This embodiment uses the specific application scenario of edge-finding embroidery of small stamps as an example to further explain the vision system-based embroidery control method described in the previous embodiment. In this scenario, the object to be embroidered is an irregularly shaped stamp, and its actual morphological parameters are specifically reflected in the outline of the stamp.

[0058] Specifically, in traditional stamp embroidery, workers must manually align the edges of the stamps and draw the embroidery patterns, which is not only inefficient but also highly prone to failure due to human error. This embodiment automatically identifies the outline of the stamps and generates or matches patterns using a vision system, achieving automation and intelligence in stamp embroidery. Figure 1 As shown, the specific implementation process for this application scenario is as follows: Step S701: Obtain an image of the badge to be embroidered through a vision system.

[0059] Specifically, the operator places the stamp to be embroidered on the embroidery frame of the embroidery machine. The image acquisition unit of the vision system takes a picture of the stamp, acquiring a raw image containing the stamp's edges and surface texture. To ensure image quality and avoid ambient light interference, a ring-shaped supplementary lighting device can be triggered simultaneously during the shooting process.

[0060] Step S702: Based on the image, identify the actual shape parameters of the badge to be embroidered, i.e. the outline of the badge.

[0061] Specifically, the AI ​​processing unit receives the original image and analyzes it using image processing algorithms. The system first filters out noise, fabric wrinkles, stains, and other interfering factors from the image, and then accurately extracts the edge contour coordinates of the badge. These contour coordinates are the actual shape parameters of the badge, representing its actual shape and position within the embroidery frame.

[0062] Step S703: Based on the actual shape parameters, adaptively reconstruct the embroidery pattern data so that the reconstructed pattern data matches the actual shape of the badge to be embroidered.

[0063] Specifically, the adaptive reconstruction in this step is manifested as a creative reconstruction process of pattern generation or pattern matching. Based on the identified Zhangzai outline, the AI ​​processing unit executes one of the following two logics: Firstly, the pattern generation mode. Based on the outline shape of the badge, the system automatically calculates the center line or edge line of the outline and generates entirely new embroidery pattern data according to preset stitch types and lengths. For example, the system can automatically generate edging patterns along the edge of the badge, or generate filling patterns to fill the internal area of ​​the badge. The generated pattern data perfectly matches the actual outline of the badge, eliminating the need for manual drawing.

[0064] Secondly, the pattern matching mode. The system compares the identified stamp outline features with the pre-stored pattern database, retrieves the existing pattern template that best matches the current stamp shape, and scales or rotates the template according to the actual size of the stamp to fit the current stamp.

[0065] Regardless of the mode used, the final generated pattern data is dynamically generated based on the actual outline of the pattern, rather than a fixed template. This reflects the core idea of ​​adaptive reconstruction in this invention.

[0066] Step S704: Control the embroidery execution unit to perform embroidery work according to the reconstructed embroidery pattern data.

[0067] Specifically, the embroidery execution unit receives the reconstructed pattern data sent by the AI ​​processing unit, drives the embroidery frame and needle to work together, and embroiders on the patch according to the generated trajectory. Since the pattern data is generated based on the actual outline of the patch, the embroidery stitches can accurately follow the edge or interior of the patch, avoiding embroidery deviation caused by the patch being placed skewed or having an irregular shape.

[0068] This embodiment automates the process of embroidering small stamps by following the steps described above. Operators simply place the stamps anywhere, and the system automatically identifies the outline and generates a suitable pattern. This effectively solves the problems of reliance on manual positioning, low efficiency, and poor precision in traditional techniques, significantly improving the production efficiency and finished product quality of small stamp embroidery.

[0069] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. It should be understood that those skilled in the art, after reading this specification, can easily conceive of arbitrary combinations of the technical features described in one or more embodiments, or equivalent substitutions of some of the technical features, based on the technical solutions described in the claims; and these combinations, substitutions, or variations do not depart from the spirit and scope of the technical solutions of the present invention, and should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. An embroidery control method based on a vision system, characterized in that, include: Acquire images of the object to be embroidered using a vision system; Based on the image, identify the actual shape parameters of the object to be embroidered; Based on the actual shape parameters, the embroidery pattern data is adaptively reconstructed so that the reconstructed pattern data matches the actual shape of the object to be embroidered. The control embroidery execution unit performs embroidery operations based on the reconstructed embroidery pattern data.

2. The embroidery control method based on a vision system according to claim 1, characterized in that, The actual morphological parameters include at least one of the outline of the object to be embroidered, the position of feature points, or the amount of deformation.

3. The embroidery control method based on a vision system according to claim 1, characterized in that, The adaptive reconstruction includes at least one of coordinate transformation, size scaling, or stitch density adjustment of the embroidery pattern data.

4. The embroidery control method based on a vision system according to claim 3, characterized in that, The pin density adjustment includes: Based on the amount of deformation in the actual shape parameters, the stitch step length is adjusted so that the stitch density matches the degree of deformation of the object to be embroidered.

5. The embroidery control method based on a vision system according to claim 2, characterized in that, The deformation amount is obtained in the following way: Obtain the standard template image corresponding to the object to be embroidered; The deformation amount is calculated by comparing the image of the object to be embroidered with the standard template image at the pixel level.

6. The embroidery control method based on a vision system according to claim 2, characterized in that, The actual morphological parameters are the feature point positions of the object to be embroidered. Based on the location of the feature points, the actual embroidery position of the embroidery pattern data on the object to be embroidered is determined by coordinate conversion.

7. The embroidery control method based on a vision system according to claim 2, characterized in that, The actual shape parameter is the deformation amount of the object to be embroidered. Based on the deformation amount, the size ratio and stitch density of the embroidery pattern data are adaptively reconstructed so that the reconstructed pattern data matches the degree of deformation of the object to be embroidered.

8. The embroidery control method based on a vision system according to claim 1, characterized in that, Before acquiring the image of the object to be embroidered, the method further includes: The image acquisition unit is calibrated, and the calibration includes at least one of flatness calibration, distortion correction and coordinate origin calibration.

9. A vision-based embroidery control device, applicable to the vision-based embroidery control method as described in any one of claims 1-8, characterized in that, include: The image acquisition unit is configured to acquire an image of the object to be embroidered; The AI ​​processing unit is configured to identify the actual shape parameters of the object to be embroidered based on the image. Based on the actual shape parameters, the embroidery pattern data is adaptively reconstructed so that the reconstructed pattern data matches the actual shape of the object to be embroidered. The embroidery execution unit is configured to perform embroidery operations based on the reconstructed embroidery pattern data.

10. An embroidery control device based on a vision system according to claim 9, characterized in that, The AI ​​processing unit includes a processor and a memory. The memory stores a computer program, and the processor executes the computer program to realize the functions of identifying the actual shape parameters of the object to be embroidered and adaptively reconstructing the embroidery pattern data.

Images