Fabric texture evaluation and analysis system and method and storage medium

Abstract

The invention belongs to the technical field of textiles and garments, relates to a texture rating technology, is used for solving the problem that an accurate texture prediction model cannot be established only by depending on factor analysis in the prior art, and particularly relates to a fabric texture evaluation analysis system and method and a storage medium. The evaluation analysis platform is in communication connection with an attribute test module, a perception analysis module, a factor analysis extraction module, a model training module and a texture rating module; the attribute testing module is used for carrying out comprehensive physical attribute testing on the sample fabric and obtaining corresponding attribute parameters; the factor analysis and extraction module is used for extracting common factors according to physical attributes; according to the method, the multi-dimensional objective attribute data is simplified into a small number of common factors by adopting a factor analysis method, and the common factors can effectively reflect the main characteristics of the texture of the fabric by utilizing a regression equation, so that individual differences and evaluation deviations caused by purely relying on subjective scoring are avoided.

Description

Technical Field

[0001] The present invention belongs to the technical field of textiles and clothing, and relates to a texture rating technology, and specifically to a fabric texture evaluation and analysis system, method and storage medium. Background Art

[0002] The evaluation of fabric texture is of great significance in the textile and apparel industry. Traditional texture evaluation mostly relies on the subjective feelings of professionals, but this method is difficult to achieve a unified and standardized evaluation due to large individual differences. Therefore, the industry has gradually turned to establishing texture standards by measuring the physical properties of fabrics through instruments.

[0003] However, relying solely on physical properties cannot fully reflect consumers' true feelings about texture. Currently, common fabric testing methods include testing of objective physical properties such as bending, friction, compression, thermal conductivity, and stretching. These indicators can reflect some of the performance of fabrics in actual use, but there is still a certain gap with consumers' subjective texture experience. Therefore, how to effectively link these objective properties with human subjective perception to form a scientific, systematic and accurate rating model has become a research focus of the textile industry.

[0004] Factor analysis is a statistical method that extracts common factors from multiple variables by reducing dimensions to improve the explanatory power of the analysis. In the textile field, factor analysis can simplify the multi-dimensional objective data of fabric texture into a few common factors, thereby effectively extracting the core variables related to subjective feelings. However, relying solely on factor analysis is not enough to establish an accurate texture prediction model. The automatic rating of fabric texture is inefficient, and the texture prediction results are not consistent with the user's subjective experience.

[0005] In view of the above technical problems, this application proposes a solution. Summary of the invention

[0006] The purpose of the present invention is to provide a fabric texture evaluation and analysis system, method and storage medium, which, combined with machine learning algorithms, such as the XGBoost gradient boosting decision tree algorithm, can train and verify the common factors and subjective scores extracted by factor analysis to establish a more accurate texture rating model; XGBoost is widely used in nonlinear relationships and complex data modeling scenarios due to its superior performance in multidimensional data processing. The present invention adopts a method combining factor analysis and XGBoost, which can achieve accurate prediction and rating of fabric texture, thereby forming a standardized and quantitative evaluation system, solving the subjective problem of traditional texture evaluation, providing consumers with a reliable basis for purchase, and providing scientific guidance for corporate research and development and design;

[0007] The technical problem to be solved by the present invention is: how to provide a precise texture prediction model to achieve accurate prediction and rating of fabric texture.

[0008] The purpose of the present invention can be achieved through the following technical solutions:

[0009] A fabric texture evaluation and analysis system comprises an evaluation and analysis platform, wherein the evaluation and analysis platform is communicatively connected with an attribute testing module, a perception analysis module, a factor analysis and extraction module, a model training module and a texture rating module;

[0010] The property testing module is used to perform a comprehensive physical property test on the sample fabric and obtain corresponding property parameters;

[0011] The perception analysis module is used to score the texture of the fabric according to actual perception: the score is subdivided into five dimensions: softness, smoothness, elasticity, thickness and warmth;

[0012] The factor analysis and extraction module is used to extract common factors according to physical properties: converting the properties of the fabric such as bending performance, friction coefficient, compressibility, thermal conductivity and heat flow into common factors and constructing a regression equation; the common factors include bending thermal conductivity factor, compression recovery factor and friction heat flow factor;

[0013] The model training module is used to perform XGBoost model training;

[0014] The texture rating module is used to automatically rate the texture of new fabrics: the common factors of the new fabrics are calculated, and then the common factors are input into the XGBoost model to automatically predict the subjective texture score of the new fabrics.

[0015] Furthermore, the physical property test contents include bending performance test, friction coefficient test, compression test, thermal conductivity test and heat flow test; the property parameters obtained from the bending performance test include the average stiffness of the longitudinal flexural force, the average stiffness of the latitudinal flexural force, the work of the longitudinal flexural force, and the work of the latitudinal flexural force; the property parameters obtained from the friction coefficient test include the longitudinal surface friction coefficient and the latitudinal surface friction coefficient; the property parameters obtained from the compressibility test include compression work, compression average stiffness, recovery average stiffness, compression recovery rate, and thickness; the property parameters obtained from the thermal conductivity test include compression thermal conductivity and recovery thermal conductivity; the property parameters obtained from the heat flow test include maximum heat flux.

[0016] Furthermore, the texture score is obtained by textile professionals who score the fabrics based on their actual experience, and the score is assigned through the empirical assignment method to ensure the objectivity and consistency of the score.

[0017] Furthermore, the specific process of common factor extraction includes: calculating the correlation matrix between all objective attributes to determine the degree of correlation between the objective attributes of the fabric; then using the principal component analysis method to extract the common factors; then, using the orthogonal rotation method to rotate the component matrix to improve the interpretability of the common factors, thereby determining the bending thermal conductivity factor, compression recovery factor and friction heat flow factor.

[0018] Furthermore, the process of XGBoost model training includes: taking common factors and subjective scores as input items, using XGBoost gradient boosting decision tree algorithm to train the sample data set, and XGBoost algorithm improves the accuracy and stability of the model through cross-validation and hyperparameter optimization.

[0019] Furthermore, the specific training process includes:

[0020] Step K1: Data standardization: Use StandardScaler to standardize the input features so that their mean is 0 and their standard deviation is 1;

[0021] Step K2: Data preparation and segmentation: The extracted common factors and texture scores are used as the input data set of the model, where the input data set is divided into a training set and a test set, with 90% of the data used for training the model and 10% of the data used for testing the performance of the model;

[0022] Step K3: Setting up the XGBoost model: Setting the objective function and random seed to ensure the reproducibility of the results; defining the set of hyperparameters to be optimized through the hyperparameter style param_grid to find the optimal model;

[0023] Step K4: Hyperparameter optimization: GridSearchCV is used to optimize the hyperparameters, and the stability of the model is ensured by the 5-fold cross-validation method; at the same time, neg_mean_squared_error is used as the evaluation indicator to find the model with the smallest MSE under different parameter combinations.

[0024] A method for evaluating and analyzing fabric texture comprises the following steps:

[0025] Step S1: testing the physical properties of sample fabrics based on fabric touch detection;

[0026] Step S2: subjective texture evaluation of sample fabrics based on the empirical value-added method;

[0027] Step S3: Establishing a fabric texture rating model based on factor analysis and XGBoost gradient boosting decision tree;

[0028] Step S4: Automatically rate the texture of the new fabric based on the XGBoost model.

[0029] A computer storage medium stores a computer program, which, when executed by a processor, implements a method for evaluating and analyzing fabric texture.

[0030] The present invention has the following beneficial effects:

[0031] 1. Improve the objectivity and consistency of texture evaluation: Use factor analysis to simplify multidimensional objective attribute data into a few common factors, and use regression equations to ensure that these common factors can effectively reflect the main characteristics of fabric texture, thereby avoiding individual differences and evaluation bias caused by relying solely on subjective scoring;

[0032] 2. Enhance the adaptability and stability of the model: Use the XGBoost gradient boosting decision tree algorithm to train the texture rating model. Through hyperparameter optimization and cross-validation, the stability and prediction accuracy of the model are significantly improved, making it adaptable to different types of fabrics and improving the applicability of the model;

[0033] 3. Support standardized management of large-scale fabric data: The model can be integrated into the fabric texture database to conduct large-scale texture testing and automated rating of fabrics of different categories, making the storage and management of texture information more efficient. The database can be widely used on e-commerce platforms to provide consumers with fabric texture references and enhance the shopping experience.

[0034] 4. Assist enterprises in R&D and design improvement: By providing accurate data-based texture ratings, R&D personnel and designers can intuitively understand the characteristics of different fabrics, promote effective communication between teams, and facilitate targeted adjustments to designs and improvements in materials, thereby achieving upgrades and optimizations in fabric performance;

[0035] 5. The present invention has achieved a technological breakthrough in the field of fabric texture evaluation, providing the textile industry with a scientific, data-driven texture rating standard, effectively supporting multiple application scenarios such as e-commerce sales, R&D management, and product design, and has significant economic and social benefits. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0037] Figure 1 is a system block diagram of Embodiment 1 of the present invention;

[0038] Figure 2This is a flow chart of the method of Embodiment 2 of the present invention. DETAILED DESCRIPTION

[0039] The technical scheme of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0040] Embodiment 1: Figure 1 As shown, a fabric texture evaluation and analysis system includes an evaluation and analysis platform, which is communicatively connected with an attribute testing module, a perception analysis module, a factor analysis and extraction module, a model training module, and a texture rating module.

[0041] The property testing module is used to conduct a comprehensive physical property test on the sample fabric and obtain the corresponding property parameters. The physical property test contents include bending performance test, friction coefficient test, compressibility test, thermal conductivity test and heat flow test; the property parameters obtained from the bending performance test include the average stiffness of the warp bending force, the average stiffness of the weft bending force, the work of the warp bending force, and the work of the weft bending force; the property parameters obtained from the friction coefficient test include the warp surface friction coefficient and the weft surface friction coefficient; the property parameters obtained from the compressibility test include compression work, compression average stiffness, recovery average stiffness, compression recovery rate, and thickness; the property parameters obtained from the thermal conductivity test include compression thermal conductivity and recovery thermal conductivity; the property parameters obtained from the heat flow test include maximum heat flux.

[0042] The perception analysis module is used to score the texture of fabrics based on actual perception: textile professionals are invited to score the texture of fabrics based on actual experience, and the scores are subdivided into five dimensions: softness, smoothness, elasticity, thickness and warmth; the scores are assigned through the empirical assignment method to ensure that the scores have a certain degree of objectivity and consistency; a six-level scale is used for assignment, such as warmth can be divided into level 0 - very cool, level 1 - cool, level 2 - slightly cool, level 3 - moderate, level 4 - slightly warm, level 5 - hot, level 6 - very hot, a total of 7 sub-feelings, and the sub-items have 10 scales (for example, level 0 to level 1, divided into 0.1, 0.2, 0.3...0.9, 1.0). The premise of the empirical assignment method is that the evaluators need to be experienced practitioners in the textile field, such as designers, fabric developers, etc., who generally have similar or consistent perceptions of various fabric textures; after the evaluators are selected, they can just check the corresponding scale according to their actual perception.

[0043] The factor analysis extraction module is used to extract common factors based on physical properties: the properties of the fabric such as bending performance, friction coefficient, compressibility, thermal conductivity and heat flow are converted into common factors such as bending thermal conductivity factor, compression recovery factor and friction heat flow factor, and regression equations are constructed; factor analysis can significantly reduce the data dimension and extract representative common characteristics; factor analysis is a commonly used data analysis method in statistics, which can generally be operated using professional software such as SPSS. The specific process is: calculate the correlation matrix between all objective attributes to determine the degree of correlation between the objective attributes of the fabric; and then use the principal component analysis method to extract common factors (generally, the eigenvalue of the common factor is required to be greater than 1 and the cumulative contribution rate of all common factors is greater than 80%. Then, the component matrix is ​​rotated by orthogonal rotation to improve the interpretability of the common factors, so that common factors such as bending thermal conductivity factor, compression recovery factor and friction heat flow factor can be determined; the multidimensional objective attribute data is simplified into a few common factors by using factor analysis, and regression equations are used to ensure that these common factors can effectively reflect the main characteristics of fabric texture, thereby avoiding individual differences and evaluation bias caused by relying solely on subjective scoring.

[0044] The regression equation of bending thermal conductivity factor is: bending thermal conductivity factor = 0.177 × average stiffness of warp bending force + 0.160 × average stiffness of weft bending force + 0.183 × work of warp bending force + 0.163 × work of weft bending force - 0.011 × friction coefficient of warp surface - 0.079 × friction coefficient of weft surface + 0.059 × thickness + 0.034 × compression work - 0.033 × compression recovery rate - 0.043 × average stiffness of compression - 0.011 × average stiffness of recovery + 0.188 × compression thermal conductivity + 0.189 × recovery thermal conductivity + 0.045 × maximum heat flux;

[0045] The regression equation of the compression recovery factor is: compression recovery factor = 0.022 × average stiffness of warp bending force + 0.052 × average stiffness of latitudinal bending force - 0.009 × work of warp bending force + 0.027 × work of latitudinal bending force + 0.036 × coefficient of friction of warp surface + 0.101 × coefficient of friction of latitudinal surface - 0.032 × thickness - 0.273 × compression work - 0.267 × compression recovery rate + 0.271 × average stiffness of compression + 0.298 × average stiffness of recovery - 0.097 × thermal conductivity of compression - 0.071 × thermal conductivity of recovery - 0.009 × maximum heat flux;

[0046] The regression equation of the friction heat flow factor is: Friction heat flow factor = -0.027×average stiffness of longitudinal flexure + 0.048×average stiffness of latitudinal flexure-0.038×work of longitudinal flexure + 0.066×work of latitudinal flexure + 0.250×coefficient of longitudinal surface friction + 0.321×coefficient of latitudinal surface friction + 0.232×thickness + 0.030×compression work-0.128×compression recovery rate-0.027×average stiffness of compression + 0.059×average stiffness of recovery-0.081×compression thermal conductivity-0.092×recovery thermal conductivity-0.303×maximum heat flux.

[0047] The model training module is used to train the XGBoost model: common factors and subjective scores are used as input items, and the sample data set is trained using the XGBoost gradient boosting decision tree algorithm. The XGBoost algorithm ensures the accuracy and stability of the model through cross-validation and hyperparameter optimization, thereby achieving accurate prediction of the texture score; the texture rating model is trained using the XGBoost gradient boosting decision tree algorithm, and the stability and prediction accuracy of the model are significantly improved through hyperparameter optimization and cross-validation, so that it can adapt to different types of fabrics.

[0048] The specific training process includes the following steps:

[0049] Step K1: Data standardization: Use StandardScaler to standardize the input features so that their mean is 0 and their standard deviation is 1, ensuring that the feature data is on the same scale and avoiding excessive impact of certain features on the model; StandardScaler is a preprocessing tool in the scikit-learn library for data standardization; Standardization is the process of scaling features to a given range (usually a mean of 0 and a standard deviation of 1), and StandardScaler achieves this conversion by calculating the mean and standard deviation of the feature column.

[0050] Step K2: Data preparation and segmentation: The extracted common factors and subjective texture scoring model are used as input data sets, where the data sets are divided into training sets and test sets, with 90% of the data used for training the model and 10% of the data used for testing the performance of the model;

[0051] Step K3: Setting up the XGBoost model: Setting the objective function (squared error reg:squarederror) and random seed to ensure the reproducibility of the results; defining the set of hyperparameters to be optimized, such as the number of trees, learning rate, tree depth, regularization parameters, etc., through the hyperparameter style param_grid to find the optimal model;

[0052] Step K4: Hyperparameter optimization: GridSearchCV is used to optimize the hyperparameters, and the stability of the model is ensured by the 5-fold cross validation method. At the same time, neg_mean_squared_error is used as the evaluation indicator to find the model with the smallest MSE under different parameter combinations. The 5-fold cross validation method is a commonly used machine learning model evaluation technique. By dividing the data set into five subsets for training and testing, the model performance is repeatedly evaluated and averaged to improve the reliability and generalization ability of the estimate.

[0053] This model can be integrated into the fabric texture database for large-scale texture testing and automated rating of fabrics of different categories, making the storage and management of texture information more efficient. The database can be widely used on e-commerce platforms to provide consumers with fabric texture references and enhance their shopping experience.

[0054] The texture rating module is used to automatically rate the texture of new fabrics: by calculating the common factors of new fabrics through the same physical property tests, and then inputting the common factors into the XGBoost model, the subjective texture score of the new fabric can be automatically predicted; by providing accurate digitized texture ratings, R&D personnel and designers can intuitively understand the characteristics of different fabrics, promote effective communication between teams, and facilitate targeted adjustments to designs and improvements to materials, thereby achieving upgrades and optimizations in fabric performance.

[0055] Embodiment 2: Figure 2 As shown, a method for evaluating and analyzing fabric texture comprises the following steps:

[0056] Step S1: testing the physical properties of sample fabrics based on fabric touch detection;

[0057] Step S2: subjective texture evaluation of sample fabrics based on the empirical value-added method;

[0058] Step S3: Establishing a fabric texture rating model based on factor analysis and XGBoost gradient boosting decision tree;

[0059] Step S4: Automatically rate the texture of the new fabric based on the XGBoost model.

[0060] The present invention also includes a readable storage medium on which a computer program is stored. When the program is executed by a processor, the above-mentioned fabric texture evaluation and analysis method is implemented; those skilled in the art can understand that all or part of the steps of implementing the above-mentioned method embodiments can be completed by hardware related to the computer program. The aforementioned computer program can be stored in a computer-readable storage medium. When the program is executed, the steps of the above-mentioned method embodiments are executed; and the aforementioned storage medium includes: ROM, RAM, magnetic disk or optical disk and other media that can store program codes.

[0061] A fabric texture evaluation and analysis system, method and storage medium. When working, the physical properties of sample fabrics are tested based on fabric touch detection; the subjective texture of sample fabrics is evaluated based on the empirical value-added method; a fabric texture rating model is established based on factor analysis and XGBoost gradient boosting decision tree; and the texture of new fabrics is automatically rated based on the XGBoost model.

[0062] The above contents are merely examples and explanations of the structure of the present invention. The technicians in this technical field may make various modifications or additions to the specific embodiments described or replace them in a similar manner. As long as they do not deviate from the structure of the invention or exceed the scope defined by the claims, they should all fall within the protection scope of the present invention.

[0063] In the description of this specification, the description with reference to the terms "one embodiment", "example", "specific example", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0064] The preferred embodiments of the present invention disclosed above are only used to help explain the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to only specific implementation methods. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can understand and use the present invention well. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. A fabric texture evaluation and analysis system, characterized in that: It includes an evaluation and analysis platform, which is communicatively connected with an attribute testing module, a perception analysis module, a factor analysis and extraction module, a model training module and a texture rating module; The property testing module is used to perform a comprehensive physical property test on the sample fabric and obtain corresponding property parameters; The perception analysis module is used to score the texture of the fabric according to actual perception: the score is subdivided into five dimensions: softness, smoothness, elasticity, thickness and warmth; The factor analysis and extraction module is used to extract common factors according to physical properties: the bending performance, friction coefficient, compressibility, thermal conductivity and heat flow of the fabric are converted into common factors to construct a regression equation; the common factors include bending thermal conductivity factor, compression recovery factor and friction heat flow factor; The model training module is used to perform XGBoost model training; The texture rating module is used to automatically rate the texture of new fabrics: the common factors of the new fabrics are calculated, and then the common factors are input into the XGBoost model to automatically predict the subjective texture score of the new fabrics.

2. A fabric texture evaluation and analysis system according to claim 1, characterized in that: The physical property test contents include bending performance test, friction coefficient test, compression test, thermal conductivity test and thermal flow test; the property parameters obtained from the bending performance test include the average stiffness of the warp bending force, the average stiffness of the weft bending force, the work of the warp bending force and the work of the weft bending force; The property parameters obtained from the friction coefficient test include the longitudinal surface friction coefficient and the latitudinal surface friction coefficient; the property parameters obtained from the compressibility test include compression work, compression average stiffness, recovery average stiffness, compression recovery rate, and thickness; the property parameters obtained from the thermal conductivity test include compression thermal conductivity and recovery thermal conductivity; the property parameters obtained from the heat flow test include maximum heat flux.

3. A fabric texture evaluation and analysis system according to claim 2, characterized in that: The texture score is obtained by textile professionals based on their actual experience of the fabrics. The score is assigned through the empirical assignment method to ensure objectivity and consistency.

4. A fabric texture evaluation and analysis system according to claim 3, characterized in that: The specific process of common factor extraction includes: calculating the correlation matrix between all objective attributes to determine the degree of correlation between the objective attributes of the fabric; then using the principal component analysis method to extract the common factors; then, using the orthogonal rotation method to rotate the component matrix to improve the interpretability of the common factors, thereby determining the bending thermal conductivity factor, compression recovery factor and friction heat flow factor.

5. A fabric texture evaluation and analysis system according to claim 4, characterized in that: The process of XGBoost model training includes: taking common factors and subjective scores as input items, using XGBoost gradient boosting decision tree algorithm to train sample data sets, and XGBoost algorithm improves the accuracy and stability of the model through cross-validation and hyperparameter optimization.

6. A fabric texture evaluation and analysis system according to claim 5, characterized in that: The specific training process includes: Step K1: Data standardization: Use StandardScaler to standardize the input features so that their mean is 0 and their standard deviation is 1; Step K2: Data preparation and segmentation: The extracted common factors and texture scores are used as the input data set of the model, where the input data set is divided into a training set and a test set, with 90% of the data used for training the model and 10% of the data used for testing the performance of the model; Step K3: Setting up the XGBoost model: Setting the objective function and random seed to ensure the reproducibility of the results; defining the set of hyperparameters to be optimized through the hyperparameter style param_grid to find the optimal model; Step K4: Hyperparameter optimization: GridSearchCV is used to optimize the hyperparameters, and the stability of the model is ensured by the 5-fold cross-validation method; at the same time, neg_mean_squared_error is used as the evaluation indicator to find the model with the smallest MSE under different parameter combinations.

7. A method for evaluating and analyzing fabric texture, characterized in that: The following steps are involved: Step S1: testing the physical properties of sample fabrics based on fabric touch detection; Step S2: subjective texture evaluation of sample fabrics based on the empirical value-added method; Step S3: Establishing a fabric texture rating model based on factor analysis and XGBoost gradient boosting decision tree; Step S4: Automatically rate the texture of the new fabric based on the XGBoost model.

8. A computer storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method for evaluating and analyzing fabric texture described in claim 7 is implemented.

Images