Garment fabric size stability detection method and system

Abstract

The invention discloses a garment fabric size stability detection method and system, and relates to the technical field of garment manufacturing and quality detection.According to the garment fabric size stability detection method and system, the fabric surface morphology is accurately represented by obtaining three-dimensional point cloud data, wrinkle feature points are extracted in combination with curvature calculation so as to eliminate noise interference, and the detection accuracy is improved. The non-rigid registration algorithm is adopted to process non-uniform deformation in the dry cleaning process, the quantitative size stability is calculated based on the distance change, the problems that in the prior art, manual detection repeatability is poor, feature points are misjudged through an automatic scheme, registration fails and the like are effectively solved, the microscopic wrinkle change of the fabric before and after dry cleaning is accurately captured, and the detection accuracy is improved. The registration problem caused by non-uniform deformation is effectively solved, the accuracy and reliability of dimensional stability detection are improved, and personal errors are avoided.

Description

Technical Field

[0001] This application relates to the field of garment manufacturing and quality inspection technology, and in particular to methods and systems for testing the dimensional stability of garment fabrics. Background Technology

[0002] In the field of garment manufacturing and quality inspection, fabric dimensional stability is a core indicator for measuring fabric quality, directly affecting the fit and durability of finished garments. Traditional testing methods generally rely on manual measurement, requiring operators to compare the dimensions of samples multiple times before and after dry cleaning using rulers. This process is not only time-consuming and labor-intensive, but also suffers from poor repeatability due to variations in human operation. For example, when measuring lightweight fabrics, uneven stretching can easily lead to millimeter-level errors. More importantly, manual methods cannot effectively capture the dynamic changes in microscopic wrinkles on the fabric surface, especially for easily deformable materials such as silk and knitwear. The complex wrinkle patterns produced after dry cleaning are difficult to quantify and assess with the naked eye.

[0003] While existing automated inspection solutions attempt to incorporate optical scanning technology, significant shortcomings remain in practical applications: When dealing with dry-cleaned fabrics, the uneven curvature distribution and blurred boundaries of wrinkled areas often cause existing algorithms to misclassify noise as wrinkle feature points, resulting in distorted feature point data. Furthermore, the non-uniform deformation of the fabric during dry cleaning renders rigid registration methods ineffective, making it impossible to accurately match the correspondence between wrinkles before and after dry cleaning during spatial registration. Additionally, dimensional change calculations rely solely on overall contour comparison, ignoring the impact of local wrinkle density differences on stability, leading to detection results that deviate from the actual deformation state. These issues render existing solutions unreliable when evaluating fabrics with different fiber compositions or structures, failing to meet the high-precision quality control requirements of garment production.

[0004] The above content is only used to help understand the technical solution of this application and does not represent an admission that the above content is prior art. Summary of the Invention

[0005] The main objective of this application is to provide a method and system for testing the dimensional stability of clothing fabrics, aiming to improve the accuracy and reliability of testing the dimensional stability of clothing fabrics.

[0006] To achieve the above objectives, this application proposes a method for testing the dimensional stability of clothing fabrics, the method comprising: Acquire the surface three-dimensional point cloud data of the garment fabric sample before dry cleaning and the surface three-dimensional point cloud data after dry cleaning, and use the surface three-dimensional point cloud data before dry cleaning as the reference point cloud data, and the surface three-dimensional point cloud data after dry cleaning as the processed point cloud data. Based on the baseline point cloud data and the processed point cloud data, a set of high curvature region points with curvature values ​​higher than a preset curvature threshold is extracted by curvature calculation and processing, and the pre-dry cleaning wrinkle feature point set data and the post-dry cleaning wrinkle feature point set data are generated respectively. Based on the dry cleaning post-wrinkle feature point set data and the dry cleaning pre-wrinkle feature point set data, spatial registration is performed using a non-rigid registration algorithm to generate registered point set data. Based on the registered point set data and the reference point cloud data, the size stability detection result data is generated through distance change calculation.

[0007] In one embodiment, the steps of acquiring surface three-dimensional point cloud data of a garment fabric sample before dry cleaning and surface three-dimensional point cloud data after dry cleaning, and using the surface three-dimensional point cloud data before dry cleaning as the reference point cloud data and the surface three-dimensional point cloud data after dry cleaning as the processed point cloud data include: The clothing fabric samples were pretreated in a constant temperature and humidity chamber to generate stress-stable samples. Based on the stress-stable sample, surface point information is collected using a three-dimensional scanning device to generate initial point cloud data; The initial point cloud data is subjected to noise reduction filtering to generate reference point cloud data and processed point cloud data.

[0008] In one embodiment, the step of extracting a set of high-curvature region points with curvature values ​​higher than a preset curvature threshold based on the reference point cloud data and the processed point cloud data, and generating pre-dry cleaning wrinkle feature point set data and post-dry cleaning wrinkle feature point set data respectively, includes: Calculate the local curvature value for each vertex in the reference point cloud data or the processed point cloud data to generate curvature distribution data; Based on the curvature distribution data, vertices with curvature values ​​exceeding a preset curvature threshold are selected to generate candidate feature point data. The candidate feature point data are summarized to generate the pre-dry cleaning wrinkle feature point set data and the post-dry cleaning wrinkle feature point set data.

[0009] In one embodiment, the step of generating registered point set data by spatial registration using a non-rigid registration algorithm based on the dry-cleaned wrinkle feature point set data and the dry-cleaned wrinkle feature point set data includes: Establish a point-to-point mapping relationship between the pre-dry cleaning wrinkle feature point set data and the post-dry cleaning wrinkle feature point set data, and generate feature correspondence relationship data; Based on the feature correspondence data, spatial deformation parameters are calculated using a thin plate spline transformation algorithm to generate non-rigid transformation parameters. The non-rigid transformation parameters are applied to perform displacement compensation on the dry-cleaned wrinkle feature point set data to generate the registered point set data.

[0010] In one embodiment, the step of calculating spatial deformation parameters and generating non-rigid transformation parameters based on the feature correspondence data using a thin-plate spline transformation algorithm includes: Based on the feature correspondence data, a thin plate spline energy function is constructed to generate an initial transformation model; A smoothing constraint factor is introduced to optimize the initial transformation model, generating a regularized transformation model; The non-rigid transformation parameters are generated by iteratively optimizing the solution to obtain the minimum energy of the regularized transformation model.

[0011] In one embodiment, the step of applying the non-rigid transformation parameters to perform displacement compensation on the dry-cleaned wrinkle feature point set data to generate the registered point set data includes: Based on the non-rigid transformation parameters, coordinate transformation is performed on the dry-cleaned wrinkle feature point set data to generate the point set data to be verified. Calculate the topological matching degree between the data set of points to be verified and the data set of wrinkle feature points before dry cleaning; When the topology matching degree reaches the preset matching degree threshold, the data of the point set to be verified is output as the registered point set data.

[0012] In one embodiment, the step of generating dimensional stability detection result data by distance change calculation based on the registered point set data and the reference point cloud data includes: Based on the reference point cloud data, a reference measurement point set is extracted, and based on the registered point set data, a registration measurement point set is extracted; Calculate the Euclidean distance difference between the reference measurement point set and the corresponding points in the registration measurement point set to generate distance change data; Statistical analysis is performed on the distance change data to generate the dimensional stability test result data.

[0013] In one embodiment, the method further includes: Based on the pre-dry cleaning wrinkle feature point set data, the number of feature points per unit area is calculated, and areas exceeding the preset density threshold are marked as high wrinkle density areas, while areas below the preset density threshold are marked as low wrinkle density areas, thus generating a region weight distribution map. According to the regional weight distribution map, a first correction threshold range is set for the high fold density region and a second correction threshold range is set for the low fold density region, wherein the first correction threshold range is greater than the second correction threshold range. Based on the first and second correction threshold ranges, a dynamic filtering coefficient matrix corresponding to the regional weight distribution map is generated, and the dynamic filtering coefficient matrix is ​​used to perform weighted filtering on the size stability detection result data to generate optimized size stability detection result data.

[0014] In one embodiment, the method further includes: Call the pre-built fabric material database to extract the fiber composition and weave structure data of the current garment fabric sample; Load permissible deformation range data that matches fiber composition and microstructure from a pre-stored industry standard database; The dimensional stability test results are compared with the allowable deformation range data. When the test results exceed the allowable deformation range, an early warning signal is triggered to indicate that the fabric dimensional stability is insufficient.

[0015] In addition, to achieve the above objectives, this application also proposes a garment fabric dimensional stability detection system, which includes: a memory, a processor, and a garment fabric dimensional stability detection program stored in the memory and executable on the processor. The garment fabric dimensional stability detection program is configured to implement the steps of the garment fabric dimensional stability detection method.

[0016] The proposed method and system for detecting the dimensional stability of clothing fabrics in this application accurately characterizes the surface morphology of the fabric by acquiring three-dimensional point cloud data, extracts wrinkle feature points by combining curvature calculation to eliminate noise interference, uses a non-rigid registration algorithm to handle non-uniform deformation during dry cleaning, and quantifies dimensional stability based on distance changes. This effectively solves the problems of poor repeatability of manual detection, misjudgment of feature points and registration failure in automated solutions in the prior art. It achieves accurate capture of microscopic wrinkle changes before and after dry cleaning of the fabric, effectively overcomes the registration problem caused by non-uniform deformation, improves the accuracy and reliability of dimensional stability detection, and avoids human error. Attached Figure Description

[0017] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0018] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1A flowchart illustrating an embodiment of the method for testing the dimensional stability of clothing fabrics according to this application; Figure 2 For this application Figure 1 A detailed flowchart of step S100; Figure 3 For this application Figure 1 A detailed flowchart of step S200; Figure 4 For this application Figure 1 Detailed flowchart of step S300; Figure 5 For this application Figure 4 A detailed flowchart of step S320; Figure 6 For this application Figure 4 A detailed flowchart of step S330; Figure 7 For this application Figure 1 Detailed flowchart of step S400; Figure 8 A flowchart illustrating another embodiment of the method for testing the dimensional stability of clothing fabrics according to this application; Figure 9 A flowchart illustrating yet another embodiment of the method for testing the dimensional stability of clothing fabrics in this application; Figure 10 This is a structural schematic diagram of an embodiment of the garment fabric dimensional stability testing system of this application.

[0020] Explanation of icon numbers: 10. Memory; 20. Processor.

[0021] The purpose, features, and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0022] The technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of this application, but merely represents selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0023] It should be understood that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, the terms "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0024] While existing automated detection solutions for the dimensional stability of garment fabrics attempt to incorporate optical scanning technology, significant shortcomings remain in practical applications. When dealing with dry-cleaned fabrics, the uneven curvature distribution and blurred boundaries of wrinkled areas often lead existing algorithms to misclassify noise as wrinkle feature points, distorting the feature point set data. Furthermore, the non-uniform deformation of the fabric during dry cleaning renders rigid registration methods ineffective, failing to accurately match the correspondence between wrinkles before and after dry cleaning during spatial registration. Additionally, dimensional change calculations rely solely on overall contour comparison, neglecting the impact of local wrinkle density differences on stability, resulting in detection results deviating from the actual deformation state. These issues render existing solutions unreliable when evaluating fabrics with different fiber compositions or structures, making it difficult to meet the high-precision quality control requirements of garment production.

[0025] Based on this, the embodiments of this application provide a method for detecting the dimensional stability of clothing fabrics, referring to... Figure 1 The method for detecting the dimensional stability of clothing fabrics includes steps S100 to S400, wherein: Step S100: Obtain the surface three-dimensional point cloud data of the garment fabric sample before dry cleaning and the surface three-dimensional point cloud data after dry cleaning, and use the surface three-dimensional point cloud data before dry cleaning as the reference point cloud data, and use the surface three-dimensional point cloud data after dry cleaning as the processed point cloud data. Step S200: Based on the reference point cloud data and the processed point cloud data, extract the high curvature region point set with curvature values ​​higher than the preset curvature threshold through curvature calculation processing, and generate the pre-dry cleaning wrinkle feature point set data and the post-dry cleaning wrinkle feature point set data respectively. Step S300: Based on the dry-cleaned wrinkle feature point set data and the dry-cleaned wrinkle feature point set data, spatial registration is performed using a non-rigid registration algorithm to generate registered point set data. Step S400: Based on the registered point set data and the reference point cloud data, the size stability detection result data is generated through distance change calculation.

[0026] In this embodiment, acquiring the surface 3D point cloud data of the garment fabric sample before and after dry cleaning can be achieved in various ways. For example, a robotic arm equipped with a laser scanner can be used to scan the garment fabric sample from multiple angles to generate initial point cloud data; or a structured light projection device combined with a camera can be used to collect surface geometric information to form 3D point cloud data. Further, noise reduction filtering can employ bilateral filtering or Gaussian filtering algorithms to remove noise points and retain key geometric features. In this embodiment, curvature calculation is a method for extracting local geometric characteristics from the point cloud data. For example, principal component analysis can be used to calculate the local curvature value of each point, or covariance matrix decomposition can be used to evaluate the degree of curvature of each vertex in the point cloud. The selection of the preset curvature threshold can be based on empirical values ​​or determined through dynamic adjustment. Its main purpose is to filter out high-curvature point sets, thereby focusing on wrinkle feature areas. Non-rigid registration algorithms can be implemented in various ways. For example, methods based on free-form deformation models achieve spatial registration by optimizing and adjusting control points; or algorithms based on elastomer models simulate the deformation behavior of fabrics during dry cleaning, thereby establishing point-to-point mapping relationships. These methods can adapt to complex deformation scenarios and ensure registration accuracy.

[0027] This application systematically improves the objectivity and accuracy of fabric dimensional stability testing by integrating 3D point cloud data processing and non-rigid deformation analysis. Compared to traditional methods that rely on manual measurement and subjective judgment, this application acquires 3D point cloud data of the surface before and after dry cleaning and specifies the baseline and post-processing state. It then uses 3D scanning technology to capture the fabric's microscopic geometric morphology, avoiding human interference and providing a high-precision digital foundation for subsequent analysis. Furthermore, by extracting point sets from high-curvature regions based on curvature calculation, it solves the problem of wrinkle feature confusion caused by overall point cloud processing in existing technologies. Simultaneously, a non-rigid registration algorithm corrects complex deformations, significantly improving spatial alignment accuracy. Finally, based on the distance change calculation between the registered point set and the baseline point cloud, the dimensional offset is directly quantified, achieving an objective assessment of fabric dimensional stability.

[0028] The method for detecting the dimensional stability of clothing fabrics in this application first acquires three-dimensional point cloud data of the surface of the clothing fabric sample before and after dry cleaning. The data before dry cleaning is used as the baseline point cloud data, and the data after dry cleaning is used as the processed point cloud data, providing a high-precision digital foundation for subsequent analysis. The three-dimensional point cloud data can comprehensively reflect the changes in the microscopic geometric morphology of the fabric, avoiding the influence of subjective interference in traditional manual measurements, thus ensuring the objectivity and consistency of the data source.

[0029] Secondly, based on the baseline point cloud data and the processed point cloud data, a set of high-curvature regions with curvature values ​​exceeding a preset curvature threshold is extracted through curvature calculation, generating wrinkle feature point sets before and after dry cleaning, respectively. Specifically, wrinkled regions typically exhibit high local curvature characteristics, and curvature calculation can sensitively identify these bending changes. By introducing a preset curvature threshold, low-curvature noise regions are effectively filtered out, ensuring that the generated wrinkle feature point sets are precisely focused on the core wrinkle region, thus resolving the problem of wrinkle feature confusion caused by overall point cloud processing in existing technologies.

[0030] Furthermore, based on the feature point set data of wrinkles before and after dry cleaning, spatial registration is performed using a non-rigid registration algorithm to generate registered point set data. Since the dry cleaning process may cause non-uniform deformation of the fabric, such as stretching or twisting, the non-rigid registration algorithm can dynamically adapt to these complex deformations and establish a precise mapping relationship between wrinkle points. Thus, the generated registered point set data effectively corrects the local distortion problem that rigid registration cannot handle, significantly improving the accuracy of spatial alignment.

[0031] Finally, based on the registered point set data and the reference point cloud data, dimensional stability test results are generated through distance change calculation. The registered point set data ensures the reliability of the point-to-point correspondence, while the distance change calculation directly quantifies the dimensional offset of the fabric before and after dry cleaning. This processing method transforms abstract dimensional stability into measurable data, avoiding misjudgments of dimensional changes caused by registration deviations in traditional methods, thus achieving an objective and reliable assessment of fabric dimensional stability.

[0032] In summary, this technical solution, by integrating 3D point cloud data processing and non-rigid deformation analysis, systematically improves the accuracy and objectivity of fabric dimensional stability detection, and effectively solves the problems of inaccurate wrinkle feature extraction, low spatial registration accuracy, and inaccurate dimensional change calculation in traditional methods.

[0033] In one feasible implementation, refer to Figure 2 Step S100 includes steps S110 to S130, wherein: Step S110: The clothing fabric sample is pretreated in a constant temperature and humidity chamber to generate a stress-stable sample. Step S120: Based on the stress-stabilized sample, surface point information is collected using a three-dimensional scanning device to generate initial point cloud data; Step S130: Perform noise reduction filtering on the initial point cloud data to generate reference point cloud data and processed point cloud data.

[0034] In this embodiment, the constant temperature and humidity chamber is a device capable of providing a constant temperature and humidity environment. It can be implemented using a test chamber with precise temperature and humidity control functions, with the aim of eliminating residual stress inside the fabric sample. The 3D scanning device can be understood as a device used to capture the geometric information of an object's surface; specifically, it can be a laser scanner or a structured light scanner, with the aim of acquiring high-precision 3D point cloud data of the fabric surface. Noise denoising and filtering processing refers to removing noise points and outliers from the point cloud data using algorithms. This can be achieved through methods such as statistical filtering and bilateral filtering, with the aim of improving the quality and reliability of the point cloud data.

[0035] In this embodiment, the scheme first pre-treats the fabric sample in a constant temperature and humidity chamber, causing the fabric fibers to reach a stress equilibrium state under constant temperature and humidity conditions. This eliminates residual stress that may have accumulated during production or storage, ensuring that the sample dimensions are in a stable baseline state. Based on this, a 3D scanning device is used to collect surface point information from the pre-treated stress-stabilized sample, generating initial point cloud data. This process effectively avoids surface deformation errors caused by internal stress. Subsequently, the initial point cloud data undergoes noise reduction filtering to further eliminate noise and outliers introduced during the scanning process, generating high-quality baseline point cloud data and processed point cloud data. This data provides a reliable foundation for subsequent wrinkle feature extraction, spatial registration, and dimensional change calculation, significantly reducing error accumulation and thus improving the accuracy and repeatability of the dimensional stability test results. The above scheme, through the organic combination of stress elimination, precise acquisition, and data purification, not only solves the problem of baseline point cloud data distortion caused by internal fabric stress but also lays an accurate data foundation for the entire dimensional stability testing process, ensuring the objectivity and reliability of the test results.

[0036] In one feasible implementation, refer to Figure 3 Step S200 includes steps S210 to S230, wherein: Step S210: Calculate the local curvature value for each vertex in the reference point cloud data or the processed point cloud data to generate curvature distribution data; Step S220: Based on the curvature distribution data, vertices with curvature values ​​exceeding a preset curvature threshold are selected to generate candidate feature point data; Step S230: Summarize the candidate feature point data to generate the pre-dry cleaning wrinkle feature point set data and the post-dry cleaning wrinkle feature point set data.

[0037] In this embodiment, the local curvature value refers to a numerical value that characterizes the surface bending characteristics by quantifying the degree of geometric change in the neighborhood of each vertex in the point cloud. It can be implemented using methods such as principal curvature method, normal vector deviation method, or covariance matrix analysis. The purpose of introducing the local curvature value is to provide fine-grained geometric basis for subsequent extraction of high-curvature regions, thereby ensuring that real wrinkled regions can be accurately identified. The curvature distribution data is a global statistical result generated based on the local curvature values. It can be expressed through histogram analysis, probability density estimation, etc., aiming to provide basic data support for subsequent screening steps. Candidate feature point data refers to the set of vertices that meet specific conditions selected from the curvature distribution data. Its purpose is to filter out interfering points in low-curvature regions, thereby effectively suppressing the influence of scanning noise and surface defects.

[0038] In this embodiment, the above-mentioned technical solution solves the problem of inaccurate wrinkle feature point extraction through a systematic curvature analysis process. First, curvature distribution data is generated by calculating the local curvature value of each vertex in the reference point cloud data or processed point cloud data. This step quantifies the surface curvature based on the local geometry of the point cloud vertices, because real wrinkled areas inevitably exhibit significant high curvature characteristics at the microscale. The local calculation method avoids errors that may be introduced by global curvature analysis, providing a fine-grained geometric basis for feature recognition. Subsequently, based on the curvature distribution data, vertices with curvature values ​​exceeding a preset curvature threshold are selected to generate candidate feature point data. Interference points in low-curvature areas (such as flat parts of the fabric) are filtered out using a preset threshold mechanism. This effectively suppresses the influence of scanning noise and surface defects on feature extraction, ensuring that only high-curvature vertices with wrinkle features are retained in the candidate point set. Finally, the candidate feature point data are summarized to generate the pre-dry cleaning wrinkle feature point set data and the post-dry cleaning wrinkle feature point set data. The scattered candidate points are integrated into a structured feature set. This step not only preserves the topological integrity of the wrinkled area, but also establishes a comparable basis for the comparative analysis of the point sets before and after dry cleaning. It avoids the deviation caused by the missing or redundant feature points in spatial registration and dimensional change calculation, and ultimately improves the robustness of dimensional stability assessment.

[0039] In this embodiment, the above-mentioned technical solution is combined with the technique of acquiring surface three-dimensional point cloud data of clothing fabric samples before and after dry cleaning and performing noise reduction and filtering, which further optimizes the accuracy of wrinkle feature point extraction. By removing noise interference in the point cloud data preprocessing stage and combining it with the accurate calculation and screening of local curvature values, the quality of the wrinkle feature point set is significantly improved, thereby laying the foundation for subsequent non-rigid registration and dimensional stability detection.

[0040] In one feasible implementation, refer to Figure 4 Step S300 includes steps S310 to S330, wherein: Step 310: Establish a point-to-point mapping relationship between the pre-dry cleaning wrinkle feature point set data and the post-dry cleaning wrinkle feature point set data, and generate feature correspondence relationship data; Step 320: Based on the feature correspondence data, calculate the spatial deformation parameters using the thin plate spline transformation algorithm to generate non-rigid transformation parameters; Step 330: Apply the non-rigid transformation parameters to perform displacement compensation on the dry-cleaned wrinkle feature point set data to generate the registered point set data.

[0041] In this embodiment, the point-to-point mapping relationship refers to determining each pair of corresponding points in the set of wrinkle feature points before and after dry cleaning using a specific algorithm. In practical applications, this mapping relationship can be achieved through nearest neighbor search algorithms, iterative nearest point algorithms, etc., with the aim of accurately locking the deformation-sensitive positions in the set of wrinkle feature points before and after dry cleaning, thus providing a highly reliable basis for subsequent spatial registration. The thin plate spline transformation algorithm is a smooth fitting method adapted to non-rigid deformation, which can be implemented using interpolation methods based on radial basis functions or energy minimization optimization methods. The key to this algorithm is its ability to dynamically capture complex nonlinear displacement patterns between point sets, thereby generating non-rigid transformation parameters that truly reflect the local wrinkle twisting and stretching characteristics of the fabric. Displacement compensation refers to the process of coordinate correction of the wrinkle feature point set data after dry cleaning based on the non-rigid transformation parameters, which can be achieved through matrix operations or geometric transformation techniques. Its purpose is to offset the wrinkle displacement error introduced by the dry cleaning process, so that the registered point set and the point set before dry cleaning achieve precise microscale overlap in space.

[0042] In this embodiment, the above-mentioned scheme effectively solves the registration misalignment problem caused by nonlinear deformation of fabric after dry cleaning by constructing a refined process of nonrigid registration. A point-to-point mapping relationship is established between the pre-dry cleaning wrinkle feature point set data and the post-dry cleaning wrinkle feature point set data, focusing on the wrinkle feature point set rather than the full point cloud data. This avoids interference from irrelevant regions in global registration and ensures that the mapping relationship is closely related to the actual wrinkle changes. Based on this, spatial deformation parameters are calculated using a thin-plate spline transformation algorithm, directly relying on the actual displacement relationship of the wrinkle feature points rather than a pre-set rigid assumption, significantly improving the physical rationality of the deformation description. Finally, the nonrigid transformation parameters are applied to the post-dry cleaning point set, and displacement correction is performed based on the nonlinear deformation information contained in the parameters. This achieves precise spatial alignment between the registered point set and the pre-dry cleaning point set, providing data support without accumulated errors for subsequent distance change calculations. Furthermore, the above scheme, combined with the techniques for acquiring three-dimensional point cloud data of the surface before and after dry cleaning and extracting point sets from high-curvature regions, forms a complete technical chain from data acquisition to spatial registration. By focusing on the local deformation characteristics of the fold feature point set, not only is the accuracy of spatial registration improved, but the limitations of traditional methods in complex fold scenarios are also overcome, thereby significantly improving the reliability of dimensional stability detection.

[0043] In one feasible implementation, refer to Figure 5 Step S320 includes steps S321 to S323, wherein: Step S321: Construct a thin plate spline energy function based on the feature correspondence data to generate an initial transformation model; Step S322: Introduce a smoothing constraint factor to optimize the initial transformation model and generate a regularized transformation model; Step S323: The non-rigid transformation parameters are generated by iteratively optimizing the minimum energy solution of the regularized transformation model.

[0044] In this embodiment, the thin-plate spline energy function is a mathematical model used to describe the energy distribution during spatial deformation. It can be implemented using an energy calculation formula based on point-to-point mapping relationships, aiming to ensure that the initial transformation model accurately reflects the geometric distribution characteristics of the fold feature points. The smoothing constraint factor can be understood as an adjustment parameter used to suppress model oscillations caused by noise interference and sudden changes in local curvature. It can be implemented by adding a Laplace smoothing term or a Gaussian smoothing term, aiming to maintain physical rationality in densely folded regions during deformation. In practical applications, iterative optimization refers to a numerical calculation method that gradually approximates the global optimum. It can be implemented using gradient descent, Newton's method, or other numerical optimization algorithms, aiming to ensure that the energy function reaches its minimum state, thereby obtaining more accurate and robust deformation parameters.

[0045] In this embodiment, the scheme first constructs a thin-plate spline energy function based on feature correspondence data to generate an initial transformation model. This process directly relies on the actual point-to-point mapping relationship, avoiding initial modeling errors caused by data deviations. Based on this, a smoothing constraint factor is introduced to optimize the initial transformation model, generating a regularized transformation model. This effectively suppresses model oscillations caused by noise interference and sudden changes in local curvature, preventing unreasonable local distortions. Finally, the minimum energy solution of the regularized transformation model is solved through iterative optimization, generating non-rigid transformation parameters. The iterative process gradually approaches the global optimal solution, ensuring that the energy function reaches its minimum value, thus providing a highly reliable foundation for subsequent displacement compensation. Overall, the above steps, through the organic combination of energy function construction, smoothing constraint introduction, and iterative optimization, significantly enhance the stability and adaptability of the deformation model, ultimately achieving accurate quantification of complex fabric wrinkle changes. Furthermore, by combining the above technical solution with the spatial registration process of the wrinkle feature point set data before and after dry cleaning, and by optimizing the generation process of non-rigid transformation parameters, the problem of overfitting or local distortion of deformation models in high curvature regions or under noise interference is solved, thereby improving the accuracy and reliability of spatial registration.

[0046] In one feasible implementation, refer to Figure 6 Step S330 includes steps S331 to S333, wherein: Step S331: Based on the non-rigid transformation parameters, perform coordinate transformation on the dry-cleaned wrinkle feature point set data to generate the point set data to be verified. Step S332: Calculate the topological matching degree between the data set of points to be verified and the data set of wrinkle feature points before dry cleaning; Step S333: When the topology matching degree reaches the preset matching degree threshold, the data of the point set to be verified is output as the registered point set data.

[0047] In this embodiment, the data set to be verified refers to the intermediate result data generated after coordinate transformation of the dry-cleaned wrinkle feature point set data using non-rigid transformation parameters. This can be achieved through matrix operations or geometric mapping. Topological matching degree is an index used to evaluate the similarity of two sets of point sets in spatial structure. It can be achieved by calculating adjacency consistency, local geometric feature similarity, etc., with the aim of verifying the geometric continuity and structural integrity of the registered wrinkle feature point set. The preset matching degree threshold is a judgment standard dynamically set according to the fabric material characteristics. It can be set based on statistical analysis of experimental data or empirical values, with the aim of ensuring that the registration result meets the accuracy requirements of subsequent dimensional stability testing.

[0048] In this embodiment, the above-mentioned scheme effectively solves the topological mismatch problem caused by directly outputting the registration results by constructing a topological matching degree verification mechanism. Based on non-rigid transformation parameters, the coordinate transformation of the point set to be processed utilizes deformation parameters obtained from the non-rigid registration algorithm to map the post-dry-cleaning wrinkle feature point set data to the reference space, thereby generating the point set data to be verified. By calculating the topological matching degree between the point set data to be verified and the pre-dry-cleaning wrinkle feature point set data, the topological similarity of the two sets of point sets is quantitatively analyzed, and the geometric consistency of the registered wrinkle features is evaluated. When the topological matching degree reaches a preset matching degree threshold, the registration result is confirmed as valid, and the point set data to be verified is output as the registered point set data. This process not only avoids low-quality registered data from entering the subsequent dimensional change calculation stage, but also adapts to the wrinkle characteristics of different materials by dynamically setting the matching degree threshold, thereby ensuring the reliability of the dimensional stability detection results. Furthermore, this scheme, combined with the aforementioned non-rigid registration algorithm, further improves the accuracy and robustness of spatial registration, providing support for the accurate extraction of complex wrinkle features.

[0049] In one feasible implementation, refer to Figure 7 Step S400 includes steps S410 to S430, wherein: Step S410: Extract the reference measurement point set based on the reference point cloud data, and extract the registration measurement point set based on the registered point set data; Step S420: Calculate the Euclidean distance difference between the reference measurement point set and the corresponding points in the registration measurement point set, and generate distance change data; Step S430: Perform statistical analysis on the distance change data to generate the size stability detection result data.

[0050] In this embodiment, the reference measurement point set refers to a representative set of points selected from the reference point cloud data. This set can be achieved using uniform sampling, random sampling, or curvature-based filtering, aiming to ensure that the measurement points comprehensively reflect the overall morphological characteristics of the fabric. The registration measurement point set refers to the set of points extracted from the registered point set data that correspond to the reference measurement point set. This set can be achieved using a nearest neighbor search algorithm or a spatial index-based matching method, aiming to ensure the accuracy of the point-to-point mapping relationship. The distance change data refers to the data set obtained by calculating the Euclidean distance difference between corresponding points in the reference measurement point set and the registration measurement point set. This can be achieved using point-by-point calculation or batch matrix operations, aiming to quantify the specific numerical value of local dimensional changes. Statistical analysis refers to calculating the mean, variance, or other statistical indicators of the distance change data. This can be achieved using classical statistical methods or weighted average methods, aiming to suppress random errors and generate statistically significant stability indicators.

[0051] In this embodiment, the structured distance change calculation process significantly improves the accuracy and reliability of dimensional stability detection. Extracting a benchmark measurement point set from the benchmark point cloud data and then extracting a registered measurement point set from the registered point set data ensures that the selection of measurement points strictly corresponds to the spatial mapping relationship after registration. This avoids measurement deviations caused by noise in folded areas or interference from non-critical areas when directly using the full point cloud, thus laying a high-precision point-to-point foundation for subsequent calculations. Utilizing Euclidean distance to directly quantify local dimensional changes effectively avoids the error accumulation problem caused by curvature differences or non-uniform deformation in overall point cloud comparison, ensuring that the change data truly reflects the actual deformation of the fabric. Statistical analysis of the distance change data transforms discrete distance differences into statistically significant stability indicators. This not only overcomes the defect of single measurement points being easily affected by local anomalies but also ensures that the detection results can objectively characterize the overall dimensional stability of the fabric, providing a quantifiable basis for quality assessment. Furthermore, during implementation, this technical solution is closely integrated with the aforementioned steps of acquiring point cloud data, extracting wrinkle feature point sets, and spatial registration. Together, they solve the technical problems caused by point cloud noise interference, fuzzy matching of corresponding points, and accumulation of random errors, thereby achieving more efficient and accurate fabric dimensional stability detection.

[0052] In one feasible implementation, refer to Figure 8 The method further includes steps S510 to S530, wherein: Step S510: Based on the pre-dry cleaning wrinkle feature point set data, calculate the number of feature points per unit area, mark areas exceeding a preset density threshold as high wrinkle density areas, and mark areas below the preset density threshold as low wrinkle density areas, and generate a region weight distribution map. Step S520: According to the regional weight distribution map, a first correction threshold range is set for the high fold density region and a second correction threshold range is set for the low fold density region, wherein the first correction threshold range is greater than the second correction threshold range. Step S530: Based on the first correction threshold range and the second correction threshold range, generate a dynamic filtering coefficient matrix corresponding to the regional weight distribution map, and use the dynamic filtering coefficient matrix to perform weighted filtering on the size stability detection result data to generate optimized size stability detection result data.

[0053] In this embodiment, the number of feature points per unit area refers to the number of wrinkle feature points counted within a specific region, which can be achieved through grid partitioning or sliding window techniques. The purpose of this step is to quantify the density of wrinkle distribution, thus providing a basis for subsequent region division. The region weight distribution map is a spatial distribution model that intuitively characterizes the sensitivity of different regions to fabric dimensional stability by associating wrinkle density with spatial location. Its purpose is to distinguish between high-wrinkle-density and low-wrinkle-density areas for differentiated processing. The first and second correction threshold ranges are set for high-wrinkle-density and low-wrinkle-density areas, respectively, to avoid the risk of misjudgment caused by traditional uniform thresholds through differentiated threshold strategies. The dynamic filter coefficient matrix is ​​an adaptive adjustment tool that dynamically adjusts the filtering intensity according to the region weight distribution. Its purpose is to retain true deformation information while suppressing noise interference, thereby improving the accuracy of the detection results.

[0054] In this embodiment, the above-mentioned scheme effectively solves the detection bias problem caused by uneven wrinkle distribution by introducing regional weight distribution and dynamic filtering mechanisms. The number of feature points per unit area is calculated based on the pre-dry cleaning wrinkle feature point set data. This process utilizes the pre-dry cleaning wrinkle distribution as a key basis for predicting deformation after dry cleaning, because the pre-dry cleaning wrinkle features directly reflect the initial weaknesses of the fabric structure, and these areas are more prone to significant dimensional changes during dry cleaning. High-wrinkle-density areas and low-wrinkle-density areas are distinguished by threshold filtering, and a regional weight distribution map is generated. This map can intuitively represent the sensitivity of different areas to fabric dimensional stability. Subsequently, a wider first correction threshold range is set for high-wrinkle-density areas, and a narrower second correction threshold range is set for low-wrinkle-density areas. This differentiated setting stems from the inherent structural instability of high-wrinkle areas, allowing for more lenient deformation assessments to avoid misjudging normal fluctuations as defects, while low-wrinkle areas require stricter thresholds to ensure that minor deformations are not ignored. Finally, a dynamic filtering coefficient matrix is ​​generated based on the correction threshold range, and the dimensional stability detection results are weighted and filtered. This matrix dynamically adjusts the filtering intensity according to the regional weight distribution, applying a lower filtering intensity in high-wrinkle-density areas to retain true deformation information, and a higher filtering intensity in low-wrinkle-density areas to suppress noise interference. This ensures that the optimized dimensional stability detection results more accurately reflect the actual deformation state of the fabric. By taking into account the spatial differences in wrinkle distribution, accurate assessment of the actual deformation in different areas is achieved, significantly improving the reliability of the detection results, especially in densely wrinkled areas.

[0055] In one feasible implementation, refer to Figure 9 The method further includes steps S610 to S630, wherein: Step S610: Call the pre-built fabric material database to extract the fiber composition and weave structure data of the current garment fabric sample; Step S620: Load allowable deformation range data matching fiber composition and microstructure from a pre-stored industry standard database; Step S630: Compare the dimensional stability test result data with the allowable deformation range data. When the test result exceeds the allowable deformation range, trigger an early warning signal to indicate that the fabric dimensional stability is insufficient.

[0056] In this embodiment, the fabric material database refers to a dataset storing fiber composition and weave structure information for various garment fabric samples. It can be implemented using a relational or non-relational database, and its purpose is to provide accurate material information for subsequent standard matching. Fiber composition and weave structure data can be understood as key parameters describing the physical properties and microstructure of the fabric. Specifically, it can include fiber types such as cotton and polyester and their proportions, as well as fabric weave patterns such as plain weave and twill weave. Its purpose is to ensure that the material characteristic identification process is efficient and repeatable. In practical applications, the industry standard database refers to a standardized dataset containing the allowable deformation range of different fabric materials under specific conditions. It can be maintained by periodically updating industry specifications or company-defined standards. Its purpose is to dynamically adapt to the allowable deformation range of different materials, thereby improving the accuracy and applicability of the judgment.

[0057] In this embodiment, the above-mentioned technical solution achieves automated evaluation of dimensional stability test results by integrating a material database with a dynamic correlation mechanism of industry standards. First, a pre-built fabric material database is used to extract fiber composition and weave structure data. This process ensures accurate acquisition of material information and avoids errors that may be introduced by manual input. Based on this, allowable deformation range data matching the fiber composition and weave structure is loaded from the industry standard database. This step dynamically obtains applicable standards based on the extracted material characteristics, overcoming the incompatibility of general standards with different fabrics (such as cotton and synthetic fibers), making the deformation range assessment closely aligned with actual material requirements. Finally, the dimensional stability test results are compared with the allowable deformation range data and an early warning is triggered. This logic establishes an automatic judgment mechanism; when the test result exceeds a threshold, an immediate warning of insufficient stability is issued, reducing manual review steps, enhancing the practical value of the test results, and eliminating subjective judgment bias through an objective threshold mechanism. Furthermore, this solution, combined with the aforementioned wrinkle feature extraction, spatial registration, and distance change calculation processing, further improves the reliability and efficiency of the overall testing process and solves the problem of inaccurate judgment caused by differences in fabric materials.

[0058] In the embodiments of this application, the method for detecting the dimensional stability of clothing fabrics accurately characterizes the surface morphology of the fabric by acquiring three-dimensional point cloud data, extracts wrinkle feature points by combining curvature calculation to eliminate noise interference, uses a non-rigid registration algorithm to handle non-uniform deformation during dry cleaning, and calculates and quantifies dimensional stability based on distance changes. This effectively solves the problems of poor repeatability of manual detection, misjudgment of feature points and registration failure in automated solutions in the prior art, realizes the accurate capture of micro-wrinkle changes of fabrics before and after dry cleaning, effectively overcomes the registration problem caused by non-uniform deformation, improves the accuracy and reliability of dimensional stability detection, and avoids human error.

[0059] It should be noted that the above examples are only for understanding this application and do not constitute a limitation on the method for testing the dimensional stability of clothing fabrics in this application. Any simple modifications based on this technical concept are within the scope of protection of this application.

[0060] This application also provides a system for detecting the dimensional stability of clothing fabrics, for reference. Figure 10 The garment fabric dimensional stability detection system includes: a memory 10, a processor 20, and a garment fabric dimensional stability detection program stored on the memory 10 and executable on the processor 20. The garment fabric dimensional stability detection program is configured to implement the steps of the garment fabric dimensional stability detection method.

[0061] The garment fabric dimensional stability testing system provided in this application, employing the garment fabric dimensional stability testing method described in the above embodiments, can improve the accuracy and reliability of garment fabric dimensional stability testing. Compared with the prior art, the beneficial effects of the garment fabric dimensional stability testing system provided in this application are the same as those of the garment fabric dimensional stability testing method provided in the above embodiments, and other technical features of the garment fabric dimensional stability testing system are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0062] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0063] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. All equivalent structural transformations made under the technical concept of this application using the contents of the specification and drawings of this application, or direct / indirect applications in other related technical fields, are included within the scope of patent protection of this application.

Claims

1. A method for testing the dimensional stability of clothing fabrics, characterized in that, The method includes: Acquire the surface three-dimensional point cloud data of the garment fabric sample before dry cleaning and the surface three-dimensional point cloud data after dry cleaning, and use the surface three-dimensional point cloud data before dry cleaning as the reference point cloud data, and the surface three-dimensional point cloud data after dry cleaning as the processed point cloud data. Based on the baseline point cloud data and the processed point cloud data, a set of high curvature region points with curvature values ​​higher than a preset curvature threshold is extracted by curvature calculation and processing, and the pre-dry cleaning wrinkle feature point set data and the post-dry cleaning wrinkle feature point set data are generated respectively. Based on the dry cleaning post-wrinkle feature point set data and the dry cleaning pre-wrinkle feature point set data, spatial registration is performed using a non-rigid registration algorithm to generate registered point set data. Based on the registered point set data and the reference point cloud data, the size stability detection result data is generated through distance change calculation.

2. The method of claim 1, wherein the fabric is a garment fabric. The steps for acquiring surface 3D point cloud data of garment fabric samples before and after dry cleaning, and using the surface 3D point cloud data before dry cleaning as the baseline point cloud data and the surface 3D point cloud data after dry cleaning as the processed point cloud data, include: The clothing fabric samples were pretreated in a constant temperature and humidity chamber to generate stress-stable samples. Based on the stress-stable sample, surface point information is collected using a three-dimensional scanning device to generate initial point cloud data; The initial point cloud data is subjected to noise reduction filtering to generate reference point cloud data and processed point cloud data.

3. The method of claim 1, wherein the fabric is a garment fabric. Based on the baseline point cloud data and the processed point cloud data, the steps of extracting high curvature region point sets with curvature values ​​higher than a preset curvature threshold through curvature calculation and processing, and generating pre-dry cleaning wrinkle feature point set data and post-dry cleaning wrinkle feature point set data respectively, include: Calculate the local curvature value for each vertex in the reference point cloud data or the processed point cloud data to generate curvature distribution data; Based on the curvature distribution data, vertices with curvature values ​​exceeding a preset curvature threshold are selected to generate candidate feature point data. The candidate feature point data are summarized to generate the pre-dry cleaning wrinkle feature point set data and the post-dry cleaning wrinkle feature point set data.

4. The method of claim 1, wherein the fabric is a garment fabric. The steps for generating registered point set data by spatial registration using a non-rigid registration algorithm based on the post-dry cleaning wrinkle feature point set data and the pre-dry cleaning wrinkle feature point set data include: Establish a point-to-point mapping relationship between the pre-dry cleaning wrinkle feature point set data and the post-dry cleaning wrinkle feature point set data, and generate feature correspondence relationship data; Based on the feature correspondence data, spatial deformation parameters are calculated using a thin plate spline transformation algorithm to generate non-rigid transformation parameters. The non-rigid transformation parameters are applied to perform displacement compensation on the dry-cleaned wrinkle feature point set data to generate the registered point set data.

5. The method of claim 4, wherein the fabric is a garment fabric. Based on the aforementioned feature correspondence data, the steps for calculating spatial deformation parameters and generating non-rigid transformation parameters using a thin-plate spline transformation algorithm include: Based on the feature correspondence data, a thin plate spline energy function is constructed to generate an initial transformation model; A smoothing constraint factor is introduced to optimize the initial transformation model, generating a regularized transformation model; The non-rigid transformation parameters are generated by iteratively optimizing the solution to obtain the minimum energy of the regularized transformation model.

6. The method for testing the dimensional stability of clothing fabrics as described in claim 4, characterized in that, The steps for applying the non-rigid transformation parameters to perform displacement compensation on the dry-cleaned wrinkle feature point set data to generate the registered point set data include: Based on the non-rigid transformation parameters, coordinate transformation is performed on the dry-cleaned wrinkle feature point set data to generate the point set data to be verified. Calculate the topological matching degree between the data set of points to be verified and the data set of wrinkle feature points before dry cleaning; When the topology matching degree reaches the preset matching degree threshold, the data of the point set to be verified is output as the registered point set data.

7. The method of claim 1, wherein the fabric is a garment fabric. The steps for generating dimensional stability detection result data based on the registered point set data and the reference point cloud data through distance change calculation include: Based on the reference point cloud data, a reference measurement point set is extracted, and based on the registered point set data, a registration measurement point set is extracted; Calculate the Euclidean distance difference between the reference measurement point set and the corresponding points in the registration measurement point set to generate distance change data; Statistical analysis is performed on the distance change data to generate the dimensional stability test result data.

8. The method of claim 1, wherein the fabric is a garment fabric. The method further includes: Based on the pre-dry cleaning wrinkle feature point set data, the number of feature points per unit area is calculated, and areas exceeding the preset density threshold are marked as high wrinkle density areas, while areas below the preset density threshold are marked as low wrinkle density areas, thus generating a region weight distribution map. According to the regional weight distribution map, a first correction threshold range is set for the high fold density region and a second correction threshold range is set for the low fold density region, wherein the first correction threshold range is greater than the second correction threshold range. Based on the first and second correction threshold ranges, a dynamic filtering coefficient matrix corresponding to the regional weight distribution map is generated, and the dynamic filtering coefficient matrix is ​​used to perform weighted filtering on the size stability detection result data to generate optimized size stability detection result data.

9. The method of claim 1, wherein the fabric is a garment fabric. The method further includes: Call the pre-built fabric material database to extract the fiber composition and weave structure data of the current garment fabric sample; Load permissible deformation range data that matches fiber composition and microstructure from a pre-stored industry standard database; The dimensional stability test results are compared with the allowable deformation range data. When the test results exceed the allowable deformation range, an early warning signal is triggered to indicate that the fabric dimensional stability is insufficient.

10. A system for detecting dimensional stability of a garment fabric, the system comprising: The garment fabric dimensional stability testing system includes: a memory, a processor, and a garment fabric dimensional stability testing program stored in the memory and executable on the processor, wherein the garment fabric dimensional stability testing program is configured to implement the steps of the garment fabric dimensional stability testing method as described in any one of claims 1 to 9.

Images