A SYSTEM AND METHOD FOR DETERMINING SURFACE DEFORMATIONS AND ROUGHNESS OF FABRICS.

Abstract

This invention presents a device and method for the objective analysis of surface defects such as pilling, linting, wrinkle resistance, and dulling in textile fabric samples using image processing and artificial intelligence techniques applied to multi-angle and multi-light images obtained before and after textile fabric usage simulation; the device consists of a rotating platform, an illumination unit, an image acquisition unit, an image processing unit, and an evaluation unit. Furthermore, a method for evaluating surface changes occurring on textile fabric samples before and after the simulation process is also described.The method consists of steps including taking a fabric sample, capturing images of the fabric sample from different angles and lighting conditions, determining the fabric type through an artificial intelligence model, applying a simulation process to the fabric sample, comparing the surface properties of the sample before and after the simulation process, and determining a fabric quality score according to specific calculation methods. Figure 1

Description

1 TARIFF DETERMINING SURFACE DEFORMATIONS AND ROUGHNESS OF FABRICS A SYSTEM AND METHOD FOR TECHNICAL FIELD This invention allows for the processing of fabric surfaces through image processing technology and artificial intelligence applications. to determine their deformations and roughness more objectively and reliably It is related to a device and a method. 10 CURRENT TECHNOLOGY Pilling, linting, dulling (also known as loss of shine or fading) and fabric Fabric surface deformations such as wrinkling along with surface roughness, 15 of the fabrics This leads to deterioration and loss of visual quality. Fabric quality and lifespan are essential. These types of deformations can be detected through tests conducted with devices that simulate usage conditions. This is determined by recreating it in a laboratory environment. After the tests, the deformed The fabrics are subjected to a subjective visual examination. The degree of deformation is assessed on each surface. It is evaluated using scales consisting of black-and-white papers specific to the defect. Color, 20 All fabrics with patterned, textured or dimensional surfaces are identified through visual inspection. This method helps prevent potential inaccurate ratings. It is evaluated. US6842532B2, especially regarding wrinkle formation, depends on the fabric / textile structure and garment type. a method for three-dimensional measurement, evaluation and classification of appearance and it is related to the device. In this method, at least two, preferably four, parallel beams of light are directed at different angles. The light is directed at the fabric sample, and the light reflected from the sample surface is projected onto the fabric. It is captured in multiple images by a fixed digital camera. This multi-angle The images are analyzed to calculate surface gradients; the wrinkling on the fabric surface is 30. The quantity is determined using the parameters P and Q derived from these gradients, and the value of P+Q is... It is used as a numerical quality score representing the appearance of the fabric surface. This numerical The value is subjective, allowing for an objective grading of the fabric's appearance. It is calibrated according to the evaluation results. Accordingly, parallel light beams are viewed from different angles. light is directed onto a fabric sample; the light reflected from the surface of the sample is then directed at a stationary 35 Multiple images are captured using a digital camera; these multi-angle images are then analyzed. 2 The amount of wrinkling on the fabric surface is determined based on surface gradients, and P The P+Q value, derived from the parameters P and Q, represents the appearance of the fabric surface. It is used as a numerical quality score. DE4132992A1 is a 5-digit code for the objective evaluation of textile surface properties. It relates to the method and system. In this method, a camera is used under suitable lighting conditions. The fabric image captured is digitized, and grayscale or color rendering is applied from the image data. Statistical parameters are calculated based on density distributions. The analysis results are standardized. Reference values ​​or reference fabric images (e.g., AATCC wrinkle standards, colorless) (fabric references or initial samples) are evaluated by comparison, and these 10 The comparison results include wrinkling value, dyeing homogeneity, or color impression. Quantitative quality metrics such as variation are calculated using formulas. On the textile surface visual changes, quantitative and repeatable quality without the need for subjective evaluations. They can be expressed as scores. EP3588434B1 is used to detect fiber damage on the surface of fabric samples. imaging the same surface area of ​​the sample under different light directions and this multi-angle It involves the process of capturing images using a three-dimensional camera system. The resulting images quantify surface irregularities by applying a contrast function. 20 These parameters are converted into measurable parameters that indicate the degree of fiber damage. The generated numerical data are analyzed using statistical methods, and each fabric is examined. They are used as numerical quality indicators showing the level of degradation of the sample; this is the current a system with higher discriminatory power compared to visual rating systems Provides evaluation. 25 Current pilling assessment systems widely used in the textile industry, It is based on assessments made by the technical staff working in the laboratory. The fabric and evaluation sheets are placed in a cabinet, and the level of pilling on the fabric is measured. They are rated according to a scoring scale. While these evaluations are subjective, there are 30 different When evaluating fabric types, a lack of knowledge and experience often leads to inaccurate and inconsistent decisions. This leads to inaccurate results. Consequently, the levels of pilling in fabrics are incorrect. This determination can lead to customers being misled about fabric quality and complaints. This is the reason. The only product competing with this traditional method is SDL Atlas's "Pill Grade". This product requires the fabric to be twisted during the evaluation process, therefore, every 35 It cannot provide accurate results for the fabric type and has low repeatability. 3 Current methods do not objectively determine surface deformations of fabric samples. Determining fabric types through examination, artificial intelligence based on the obtained images. Using this method, identify the type of fabric sample and create an image based on the identified fabric type. Selecting and applying processing steps, fabric aging using a comparison algorithm. processing the pre- and post-simulation images of the sample subjected to simulation and 5 using an algorithm to eliminate the static visual effects resulting from the fabric's natural texture. It lacks the ability to lift. All the problems mentioned above ultimately necessitate an innovation in the relevant technical field. It has made it necessary. 10 A BRIEF DESCRIPTION OF THE INVENTION To eliminate the disadvantages mentioned above and to bring new advantages to the relevant technical field. In order to provide this, the present invention provides multi-angle 15 obtained before and after simulation. and textiles using image processing and artificial intelligence techniques applied to highly illuminated images. Objective assessment of surface defects such as pilling, linting, and dulling in fabric samples. It relates to a system and method for analysis. The system consists of a rotating platform, a lighting unit, an image acquisition unit, and an image processing unit. It consists of a unit and an evaluation unit; this allows the fabric sample to be examined from different angles. The fabric is illuminated and visualized, and the fabric type is automatically identified from these images. Classification becomes possible. Pre- and post-simulation images show the natural appearance of the fabric. suppressing the fixed visual effects caused by its texture and thus visually reducing surface degradation. It is processed using a reference comparison algorithm that isolates the traces; 25 according to fabric type. Following the selected image processing steps, the degree of degradation is expressed as a numerical quality score. It is calculated. Thanks to this structure, it differs from traditional visual and subjective evaluation methods. independently, on fabrics with different colors, patterns and surface characteristics A repeatable and standards-compliant quality assessment is carried out. One of the aims of the invention is to improve fabric resistance to pilling, linting, dulling and wrinkling. The aim is to ensure objective monitoring of deformations. Another aim of the invention is to preserve fabric without being affected by fabric color, pattern and surface pilling. The aim is to enable more reliable monitoring of surface deformations. 35 4 Another aim of the invention is to deform the fabric surface without damaging the fabric surface. to carry out inspections. Another aim of the invention is to provide information, at least regarding the degree of hairiness and the total number of hairs. The goal is to conduct detailed fabric quality control by collecting data. 5 Another objective of the invention is to obtain fabric surface deformation control through fabric surface analysis. Digital visual representations of fabric surfaces and fabric texture characteristics according to their topography. The goal is to enable its transfer to another medium. To achieve all the objectives stated above and as will be understood from the detailed explanation below. For this, the present invention is a system for evaluating changes in the surface properties of fabrics. It must be a frame containing at least one fabric sample, and the fabric sample must be 360 ​​degrees... It was configured to allow it to be moved to different angular positions within the range. a raised platform; designed to apply light to a fabric sample from multiple directions and angles. a structured lighting unit; various before and after at least one simulation process. We recorded images of the fabric sample at various angles and under different lighting conditions. An image acquisition system structured to detect and create multi-angle image sets. The unit is trained to identify fabric type using a pre-simulation image set, and Simulation pre- It processes the image sets after the simulation and uses a reference algorithm to determine the specified parameters. an AI structured to apply preprocessing steps depending on the type of tool. an image processing unit configured to process image sets using data, work- Using images that have gone through the processing stages, the presence and intensity of hair growth can be determined. It provides a rating system and a structured method to create a numerical quality score. The evaluation unit; the image collection unit in question will store the data. can be written and by the image processing unit and the evaluation unit in question. non-volatile, readable storage for data structured for reading. a computer-readable memory; and the platform in question, the lighting unit and the g- A device with a housing designed to contain the image acquisition unit; an access element located on the body that provides access to the components inside the body includes and the system must perform at least one simulation process through a simulation unit. Before and after, textile fabric samples found on the platform in question were created. surface changes are observed in the multi-angle images obtained by the image acquisition unit. by analyzing the sample fabric through the image processing unit and using an artificial intelligence model. 35 structuring it in such a way as to identify the types and evaluate them through an evaluation unit. It is characterized by... A possible configuration of the invention involves the artificial intelligence model processing the knitting, do- of the fabric sample. It is characterized by being structured in a way that determines whether it is made of fabric or is specially constructed. It is done in Rize. A possible configuration of the invention involves the AI ​​model in question, before simulation. trained to identify fabric type using an image set and the fabric's natural texture. pre-simulation and post-simulation to suppress static visual features resulting from to process the image sets with at least one reference algorithm and the specified fabric It is characterized by being configured to apply preprocessing steps depending on the type. Another possible configuration of the invention is that the reference algorithm is based on the simulation results. surface protrusions caused by the hair growth that occurs between them and the formations created by these protrusions Structured to suppress the fabric's natural texture in order to isolate shading effects. It is characterized by its nature. Another possible configuration of the invention is fabric surface topology scanning and visualization. configured for image acquisition, in communication with the image acquisition unit, laser scanner or lidar. It is characterized by including one of the browsers. Another possible configuration of the invention is that the image acquisition unit could be incorporated into knitting, weaving and structured to apply pre-treatment steps to at least one of the specially formulated fabric types. It is characterized by... Another possible configuration of the invention is one where the platform is structured to be stationary. and the lighting unit, through the aforementioned circular LED lighting system A circular LED light system configured to activate light sources at different angles. It is characterized by being configured to include a lighting system. Another possible configuration of the invention is that the image acquisition unit is at least below- It is structured to apply pre-treatment steps to fabric types consisting of the following: It is characterized by its materials: knitted, woven, and specially constructed. Another possible configuration of the invention is the image during the image acquisition process. It is characterized by the receiving unit being held in a fixed position. 35 6 Another possible configuration of the invention is by an artificial intelligence model. The analysis performed determined whether the fabric type was plain, patterned, or melange. It is characterized by including the identification step. Another possible configuration of the invention is by an artificial intelligence model 5 The analysis performed determines whether the fabric type is knitted, woven, or specially constructed. It is characterized by including the step. Another possible configuration of the invention is where the image acquisition unit captures the fabric surface. laser scanners, projection-based light sources, or lidar-based scanning for scanning 10 to collect surface topographic data, of a type belonging to at least one of the systems It is characterized by its structure. BRIEF DESCRIPTIONS OF THE FIGURES Figure 1 is an illustration showing the layout and components of the proposed system. Figure 2 is an angled view showing the layout and components of the proposed system. Figure 3 shows an application of the proposed system that detects fabric wrinkles. It is a picture. Figure 4 is an illustration showing the steps of the proposed method. REFERENCE NUMBERS GIVEN IN THE FIGURES 25 100 Systems 200 Platform 300 Lighting units 400 Image acquisition units 30 600 Image processing units 700 Evaluation units 800 Devices 801 Body 35 802 Access element 7 803 Mirror 804 Wrinkle detection light unit 805 Device 900 Memory DETAILED DESCRIPTION OF THE INVENTION In this detailed explanation, the subject is not included for the sole purpose of making the subject more understandable, without any This is explained with examples without creating a limiting effect. This invention is a multi-angle imaging system specifically designed for fabric surface analysis purposes. using image processing and artificial intelligence methods, before and after the simulation process. To evaluate surface changes occurring in textile fabric samples It is related to the system (100) and method. As shown in Figure 1, a configuration of the system (100) includes a rotating platform (200), a lighting unit (300), an image acquisition unit (400), an image processing unit (600), a a device comprising an evaluation unit (700), an enclosure (801) and an access element (802) (800) and includes non-volatile, computer-readable memory (900). A configuration of the device (800) is configured on the container (801) inside it. It has a container (801) which can be accessed via an access element (802); conversely, the device (800), a platform to hold at least one fabric sample of the container (801), at least one lighting It is built to accommodate the unit (300) and at least one image acquisition unit (400). The system designed to evaluate changes in the surface properties of fabrics (100) It is a structured system (100) which will hold at least one fabric sample on it. and the fabric sample can be moved to different angular positions within a 360-degree range. It is characterized by having a rotating platform (200) structured to provide. Plat- form (200) to illuminate the fabric sample on the said platform (200). It is positioned to face at least one structured lighting unit (300). Lighting- The lighting unit (300) is designed to apply light to the fabric sample from multiple directions and angles. It is structured. In addition, the device (800) has fabric samples located on the rotating platform (200). the lighting unit (300) can be used in various angular positions and under various lighting conditions. an image acquisition unit configured to record images while being illuminated under their light 35 (400) includes. The image acquisition unit (400) also includes at least the various stages of a fabric testing process. 8 It is configured to acquire images in its surroundings and to acquire and create multi-angle image sets. It has the capacity to capture images. An example configuration of the image acquisition unit (400) is a camera. radar. The simulation unit examines surface and appearance, which are commonly studied in textile quality assessment. It can be structured in a way that will produce its flaws; these flaws may include the following: Pilling, which is the clumping of fibers on the fabric surface to form small lumps; fuzzing, In other words, an increase in loose fiber ends and the formation of fibrous protrusions on the fabric surface; dulling, chafing... The fabric surface loses its original appearance due to wear and tear or use. a change in surface appearance resulting in a change in lacquer or surface character; this- Wrinkle resistance or wrinkle appearance, i.e., the fabric's resistance to folding, compression, or deformation. subsequently losing its smooth surface appearance. In a configuration of the system (100), the image obtained by the image acquisition unit (400) An image processing unit (600) configured to process image sets is provided; this- Rada image processing unit (600) uses the pre-simulation image set to type the fabric. using an artificial intelligence model trained to determine the type of fabric It provides. The image processing unit (600) provides a stable image resulting from the natural texture of the fabric. To suppress flood features, a re-analysis of pre- and post-simulation image sets was performed. It uses a reference algorithm and applies pre-treatment steps depending on the specified fabric type. The system (100) is characterized by having a configuration that includes an evaluation unit (700); Here the evaluation unit (700) uses images that have passed through preprocessing stages. It grades the presence and severity of hair growth and provides a predetermined rating. It creates a numerical quality score based on quality assessment references. 25 A configuration of the system (100) will allow the placement of a fabric sample It includes a platform (200) structured in such a way that the image acquisition unit (400) and the platform (200), to enable the viewing of the sample at different angular positions relative to each other. It can be moved within a 360-degree range. 30 In one configuration, the image processing unit (600) is placed on the rotating platform (200). by applying a grayscale transformation to the images of the fabric sample It is structured; these images are captured by the image acquisition unit (400). Each In the image, pixel 35 shows two dominant peaks that form horizontally and vertically within the spectrum. A 2D Fourier transform can be performed to determine their base positions. Image acquisition unit Derived from the distance between the fabric sample on platform (200) and (400) 9 Using a pixel / mm scale, the horizontal peak weft density (threads / cm) and the vertical peak weft density are determined. The point can provide warp density (yarns / cm). For knitted fabrics, the same spectral analysis, It can provide row frequency and bar frequency. Measurement is done by the image processing unit (600). The rotating platform (200) can be repeated in three different angular positions and the average and standard Deviation is reported. 5 In one application, the image processing unit (600) is pre-trained with fabric type labels. This will allow the same multi-angle images to be fed into a convolutional neural network (CNN). It is structured. As output, the artificial intelligence model (601) has the following fabric construction. It determines one of the classes based on probability: plain weave, straight weave, twill, satin, weft skip, warp 10 Jumper, single jersey, supreme, rib (1×1, 2×2), interlock, weft knit or warp knit. Artificial The class information processed by the intelligence model, the relevant preprocessing chain, and the wrinkle resistance values. This allows for selection. The wrinkle value here is considered normal and specific to the fabric type. It represents the wrinkle resistance values. In a configuration, the directional intensity of shadows created by surface wrinkles, Image acquisition unit under multi-angle illumination provided by lighting unit (300) (400) is used to calculate the evaluation from images obtained from the surface of the sample. unit (700), gradient magnitude, shadow impact frequency and local variance values ​​AATCC-128 It is configured to map to a replica scale (1=most wrinkled, 5=smoothest); thus, 1.0 to 20 A wrinkle resistance value ranging between 5.0 is produced. In one configuration, the image processing unit (600) processes the fabric surface in the same multi-angle grayscale. It is configured to process images by dividing them into small regional windows; during this process Average brightness and standard deviation are calculated for each window. The entire image is 25 The coefficient of variation (CV%) is calculated using the formula CV% = 100·σ / μ; furthermore, Periodic irregularity components (moiré bands, local band repeats) in the Fourier spectrum It is reported as a separate spectral disorder index. In one application, 30 uniform types of warp and weft threads with the same type and linear density are used. A rectangular sample with length L and width W is prepared from the fabric using the yarn. The mass of the sample, M, is measured using a precision balance and entered into the system. Weft density Sa and warp density are also entered. The threads / cm are measured via the image acquisition unit (400) and image processing unit (600). It is determined. The total length of the weft yarns in the sample is La, and the total length of the warp yarns is... It is denoted by Lç. The total of the warp and weft yarn crimp ratios is given by ca and cç, respectively. Yarn lengths are calculated using the following formulas: La = Sa · L · W · (1 + ca) Lç = Sç · L · W · (1 + cç) 5 Since the linear density of the yarn used in the warp and weft directions is the same, it has a common linear density. Density is denoted by T. When the value of T is expressed in tex (g / 1000 m³), ​​it represents the mass of the sample. It is given by the following relationship: M = ((La + Lç) / 100000) T From this, the common yarn linear density is calculated as follows: 10 T = 100000 · M / (La + Lç) When the relationships relating to total yarn lengths are written in place: T = 100000 · M / {L · W · [Sa · (1 + ca) + Sc · (1 + cç)]} If the fabric weight G (g / m²) is known, then the sample area is calculated as M = G · L · W / 10000. It simplifies and the common linear density is obtained by the following relationship: 15 T = 10 D / [Hr (1 + ca) + Sc (1 + cç)] Approximate linear density in applications where the curvature ratios can be neglected: T ≈ 10 · G / (Sa + Sç) The British cotton number, Ne, is obtained from the calculated tex value through the following conversion: Ne = 590.54 / T 20 This calculation uses a uniform yarn with the same linear density in both warp and weft directions. It is applied to fabrics. The system uses a single T value for the common yarn instead of separate values ​​for warp and weft. and produces a single Ne value. In an application, the image processing unit (600) and the evaluation unit (700) are produced by The outputs can be combined into a single fabric quality passport document. The document includes the fabric type, 25 warp / weft (or row / bar) density, T and Ne values ​​of the common yarn, wrinkle resistance value, Irregularity (CV% and spectral index) and fuzzing / pilling / dullness for the current batch. It includes scores; this information is presented both numerically and graphically. In a configuration, the image acquisition unit (400) is directed in various directions by the illumination unit (300). and under multiple lighting provided from different angles, the fabric is viewed from a 360-degree rotating platform. It is structured to capture images of surface wrinkles. directional density of shadows, multi-angle illumination provided by lighting unit (300). Image processing unit (600) from images obtained from the surface of the sample below It is calculated via the image processing unit (600) through this multi-angle 35 Directional intensity of shadows, gradient magnitudes, and local characteristics from illuminated images. 11 The evaluation unit (700) can be produced by calculating the variances. The invention is multi-angle lighting with local shade and shade variance calculations associated with each lighting condition It allows combining images obtained by the image processing unit (600). By processing, the invention represents 3D surface topography, fiber prominences and wrinkle depths. Using structured visual information to create a digital construct, a simple 2D 5 It can go beyond the technical benefits derived from capturing the image. As a result, invention, fiber protrusion density, bead cluster distribution, local matting areas, wrinkle direction and the capacity to analyze parameters such as depth, surface roughness, and irregularity. This technical approach can provide improvement. Image processing unit (600) or evaluation unit (700) outputs are not only for classification purposes, but also for a fabric surface 10 as a topology analysis system or, to some extent, as an optical surface morphology scanner It also makes its use possible. In a configuration, the system (100) establishes a relationship between fabric performance and structural parameters. It is structured to create a correlation database; this includes plumage, pilling and 15 Calculating the intensity of dulling as numerical quality scores, image processing unit (600) and fabric surface properties which can be performed by the evaluation unit (700) This can be achieved by executing the invention simultaneously with its definition. In this way, the invention is both Surface performance parameters and fabric structural parameters are both on the same system. Since they can be obtained simultaneously, there are 20 differences between two datasets based on two sample fabrics. This could allow for the establishment of high-value correlations. The invention, on the same sample... It is designed to measure the following: pilling degree, fuzziness degree, matte finish, wrinkle resistance, fabric density, yarn count, fabric structure type and fabric weave type. The system (100) can process such data in real time or the same data. Since it is designed to generate data in the flow, the following 25 can be generated over time using this data. Relationships of this type can be identified: tendency to pill at certain fabric densities, certain yarns. Risk of pilling within certain number ranges, and resistance to felting in certain knitted / woven structures. sensitivity, wrinkle resistance, surface topography, and fiber behavior in fabrics Performance correlations between the structure. In this way, the system (100) has a broader scope. It can provide data for predictive quality analysis and decision support systems. 30 In a configuration, the system (100) determines whether the fabric is woven or knitted, the structure To determine the type and surface properties of the fabric, the image processing unit (600) Using an artificial intelligence model pre-trained with labels, the structural analysis of the fabric can be determined. It is structured to analyze the properties of at least one fabric. 35 It is structured in such a way as to determine its class probabilistically; this class information is then pre-processed. 12 It determines the next steps in the chain. In this way, the device (800) not only takes measurements not only that, but it also identifies the sample through preliminary analysis and then determines the appropriate analysis accordingly. It can function as an adaptive textile analysis device that chooses the method. System (100) Integrated and selectable analysis methods may include, but are not limited to, the following: The warp / weft density approach for woven fabrics and the loop / column approach for knitted fabrics. the approach, different evaluation algorithms for fabrics with fluffy or embossed surfaces And different image segmentation strategies for plain, patterned, or melange fabrics. This allows... The system (100) is an intelligent measurement system that can choose its own analysis method according to the fabric type. It is able to function. In a configuration, the system (100) analyzes the numerical image data for each measurement. non-transient, computer-readable versions of the results and structural parameters by enabling the fabric samples to be stored together in memory (900) over time. It enables the creation of a digital quality memory. Memory (900), image receiving unit (400), Simulation unit, image processing unit (600) and evaluation unit (700) with data transfer 15 and is connected to enable access to the memory (900) and other system units. This accessibility and storage of fabric data between different batches of the same fabric comparison, analysis of differences between suppliers, quality over time monitoring trends, verifying customer complaints against historical data, and R&D This can enable the creation of a reference fabric data repository for the processes. 20 In a configuration, the numerical dataset produced by the system (100) is the physical fabric a representation of fabric quality such as a digital twin that represents specific characteristics of the sample It can be stored in memory (900) in such a way as to enable its creation; thus the system (100), functions as a component of a data-driven quality management infrastructure in the textile industry 25 can see. In a configuration, the system (100) is the reference used by the image processing unit (600). Algorithms and Fourier spectrum calculations as well as artificial intelligence model (601) By using a convolutional neural network (CNN) through non-standard fabrics or operators 30 Creating a new generation reference infrastructure for fabrics that are difficult to evaluate. It can provide opportunities. In one configuration, the image processing unit (600)), the device's (800) physical reference cards learning from real fabric data and image patterns without being bound by anything 35 It can provide. As a result, the system (100) addresses the following issues that the industry has traditionally struggled with: 13 It can create a new generation digital reference infrastructure for fabric groups: patterned fabrics, multi colored fabrics, high-pile fabrics, technical textiles, coated or laminated surfaces and Functional fabrics with complex surface structures. This invention therefore has a significant impact on industry. Digital reference systems are insufficient in areas where existing reference systems, such as traditional reference cards, are inadequate. It can provide an evaluation standard. 5 In one configuration, the system (100) displays the same multi-angle grayscale images of the fabric surface. surface through image processing unit (600) that works by dividing into small regional windows to map local defects by analyzing their spatial distribution It is structured; meanwhile, the average brightness and standard deviation are calculated for each window. 10 Coefficient of variation and periodic irregularity components in the Fourier spectrum (moiré) The bands (local band repeats) can be reported as separate spectral disorder indices. In this way, areas where beading is concentrated, areas where hair is clustered, localized Mapping of dull areas, wrinkle directions and intensity in image processing This can be done by unit (600). Therefore, the invention, mapping of surface defects 15 and can be used for local quality distribution analysis. In a configuration, the system (100) is made through the image processing unit (600). calculations, artificial intelligence model processes, and density determination, weaving / knitting type to reveal the classification, yarn count and other relevant fabric surface structure data 20 evaluation unit structured for (700) comparison with at least one reference data set by doing so, to detect product verification and prevent counterfeiting. It is structured. That is, the fabric density, yarn count and fabric structure obtained with the system (100). Type data is used not only for quality assessment but also for the technical aspects of specific products. It can also be used to verify conformity to the specifications; more specifically, the invention, 25 Improved supplier verification, batch conformity checks, and compliance with specifications. Confirmation, matching checks with reference samples, and identification of incorrect fabric or construction use. This could enable the determination of fabric density, yarn density, and other factors. In another application, the fabric density achieved with the invention could be determined using the invention. Number and fabric type data, product identity verification, structure-based matching, and It can be used to identify counterfeit or defective products within specific product groups. 30 In one configuration, the system (100) uses various textile analysis modules of the artificial intelligence model. This could allow them to be trained to work together; because with the image acquisition unit (400) image processing units (600) and evaluation units (700) such as analysis infrastructure, memory It is configured to work by establishing a data connection with (900). The system has a 35 (100). This structure, developed as part of the project, is designed not only for existing parameters but also... 14 It can provide a data backbone that also serves as a foundation for future textile analysis functions. Sufficient data is collected by the image acquisition unit (600) and stored in memory (900). and invention through artificial intelligence model training, classification of surface defects, and fabric production related to Defect identification, appearance-based quality scoring, post-use deterioration It can provide advanced analytics in the areas of forecasting and product type-based performance risk analysis. 5 Therefore, the invention becomes a modularly expandable textile artificial intelligence platform. It may have the capacity. In a configuration, the system (100) detects hair growth as a result of detailed technical analyses. Pilling, matting, wrinkle resistance, fabric density, yarn count, and fabric structure 10 It can be used to determine the type; at least for these detailed technical analyses as well. It can enable: performing numerical modeling of fabric surface topology, surface Establishing correlations between performance and structural parameters is an adaptive analysis task. to create the stream and build a digital quality memory / digital twin-like data structure, locally to perform defect mapping, provide technical comparisons for product validation and 15 To lay the groundwork for future textile analysis modules. In a configuration, the image processing unit (600) within the system (100) is artificial intelligence. It may include different modules structured to use the model. These include a frequency The assignment module and its application steps are placed on a rotating platform. Grayscale transformation on images of the fabric sample obtained with the same camera. The application involves performing a 2D Fourier transform on each image, in both horizontal and vertical directions across the spectrum. Finding the pixel-based location of the two dominant peaks formed; camera-fabric Using a pixel / mm scale derived from the distance, the horizontal weft density (threads / cm) and the vertical weft density can be determined. Determining the peak warp density (threads / cm); the same spectral analysis course in knitted fabrics. and to give the rod (wale) densities; the measurement is taken at three different angular positions of the rotating platform. The repetition and reporting of the mean and standard deviation can be found. Additionally, a weave can be used. or the same multiple as the knitting type classification module and the application steps belonging to this module. Angular images are fed into a convolutional neural network (CNN) previously trained with fabric type labels. Feedable; output layer, plain, twill, satin, weft skip, warp Examples include skip knit, single jersey, rib knit (1x1, 2x2), interlock, weft knit, and warp knit. One can probabilistically identify one of the classes, and the class information can trace the relevant preprocessing chain and convolution. You can select the (crimp) coefficients. Additionally, there is a wrinkle resistance module and its corresponding implementation. As a step-by-step process, in the images obtained from the surface of the sample under multi-angle illumination, The directional intensity of shadows formed by surface folds can be calculated, and the gradient can be determined. 35 Size, shadow-stroke frequency, and local variance values ​​were compared to the AATCC-128 replica scale (1=en By pairing wrinkled (5=smoothest), a decimal wrinkle-free score between 1.0 and 5.0 can be generated. Furthermore, the unevenness modulus and the application steps related to this modulus are considered as the fabric surface. The same multi-angle grayscale images are divided into small regional windows; in each window Average brightness and standard deviation are calculated; coefficient of variation across the entire image scale. CV% is derived using the formula CV% = 100·σ / μ and also the periodic irregularity in the Fourier spectrum. components (moiré bands, local band repeats) as a separate spectral irregularity index It is reported. Furthermore, in an application related to the yarn count determination module, the same type is used in warp and weft systems. and fabric made using a single type of yarn with the same linear density, length L and width W A rectangular sample can be prepared. The mass of the sample, M, is measured using a precision balance. The system is entered; the weft density Sa and warp density Sc are entered via the density determination module, yarns / cm. It is determined in terms of the total lengths of the warp and weft yarns, respectively La = Sa · L · W · (1 + The common yarn linear density T, M = (ca) and Lç = Sç · L · W · (1 + cç). From the relation ((La + Lç) / 100000) · T, we obtain T = 100000 · M / (La + Lç). The fabric... If the weight G is known, the relation T = 10 · G / [Sa · (1 + ca) + Sç · (1 + cç)] can be used. Cotton number Ne is calculated using the conversion 590.54 / T, and a single T and single T belonging to the common yarn. A value is reported. There is also a reporting module and a process implemented within that module. In this application, the outputs of all the aforementioned modules are included in a single fabric quality passport document. The following are combined in the document: fabric type, warp / weft (or row / bar) density, common T and Ne, wrinkle resistance. score, unevenness (CV% and spectral index) and in the present invention Pilling / hairiness / matte scores are presented both numerically and graphically. In another configuration of the system (100), the image acquisition unit (400) Multi-angle viewing was achieved by keeping the platform (200) fixed and a circular LED 25 a lighting system consisting of light sources at different angles through this system by using a lighting unit (300) configured to activate in sequence realizable. In another application of the system (100), data relating to surface topography, fabric surface 30 laser scanners, projection-based light sources, or lidar-based scanning for scanning. This can be obtained by an image acquisition unit (400) which includes at least one of the systems. The simulation unit performs fabric surface deformation tests in accordance with the EN ISO 12945-1 standard. Uses: Surface pilling of textile fabrics using a rotating pilling box device, 35 Assessment of susceptibility to pilling and matting; also, Martindale feathering and 16 The abrasion test conforms to the EN ISO 12945-2 standard; surface pilling of textile fabrics, Assessment of susceptibility to hairiness and dullness; also, wrinkle resistance tests, EN Complies with ISO 12945-3; prevents pilling, linting, and dulling on the surface of textile fabrics. Evaluation of the trend; also, wrinkle removal tests on textile fabrics, AATCC Test The method is in accordance with 128. The images are captured by the image acquisition unit (400) inside the device (800). The fabric samples taken are subjected to the fabric tests mentioned above by the simulation unit. It is retained and re-analyzed by the post-simulation device (800). The system shown in Figure 2 (100) shall contain at least one fabric sample. a structured platform (200); platform housing the fabric sample (200) completely 10 lighting unit containing surrounding light sources (300) and said lighting unit (300) configured to apply lighting to the fabric sample, Specifically, it will shine light at a 15-degree angle onto the upward-facing surface of the fabric sample. Light sources positioned around the platform (200) as shown; at least one simulation 15 before and after the procedure, at various angular positions and under various lighting conditions to record images of the fabric sample and create multi-angle image sets. an image acquisition unit (400) structured to provide the platform (200), lighting a housing designed to house the unit (300) and the image acquisition unit (400) (801) a device (800) and the body (801) located on the body (801) It may include an access element (802) that provides access to the components within it. The aforementioned access element 20 Access element (802) can be a cover in a possible configuration. Access element (802); cover, drawer, sliding panel, hinged panel, removable panel or access to the interior volume of the body It can be a similar structure that provides this. In a configuration of this system (100) shown in Figure 3, the device (800) contains a mirror 25 It is configured. The mirror in question is attached to the base of the inside of the device (800) body (801) 45 It can be positioned at an angle of degrees. At the same time, the fabric to be evaluated... The sample is suspended at a 5-degree angle to the access element (802) by means of a device (805). It can be positioned. Thus, the device has at least (800) wrinkle resistance in standard wrinkle resistance tests. It can be used in mode and the fabric located in the access element (802) via the mirror in question 30 The sample may be in a state where it can be imaged by the image acquisition unit (400). During wrinkle-free mode, the access element of at least one wrinkle detection light unit (804) (802) fabric positioned at a 5-degree angle on the surface remaining inside the body (801). It may be configured to shine light on the sample from above. 35 17 First angle between the mirror (803) and the base of the body (801) and access with the fabric sample. The second angle between element (802) depends on the dimensions of the device (800) and the image acquisition unit (400) the position, the dimensions of the mirror (803) and the surface of the fabric sample to be viewed This can be changed depending on the area. The first and second angles can preferably be selected between 0° and 90°. In an example application, the mirror (803) is positioned at an angle of approximately 45° to the base of the body (801); fabric 5 The specimen can be positioned at an angle of approximately 5° relative to the access element (802). With this Together, the specified angle values ​​are not limiting and the image acquisition unit (400) mirror (803) through which it can view the part of the fabric sample that is to be evaluated and Different angles that allow the identification of wrinkles on the surface of the fabric sample. The values ​​are available. 10 In one application of the method shown in Figure 4, the sample is fixed during the image acquisition process. While the camera is held in position, the lighting unit (300) is positioned from different directions and different It is configured to apply light from various angles. Light is emitted from varying angles and directions. Thanks to this, different shades can be created on the fabric surface. In this way, each lighting element can be 15... Under these conditions, the appearance of the fabric surface can be recorded by the image acquisition unit (400). In one application, the image acquisition process was performed on both the fabric sample without simulation and The same angular positions and lighting conditions were used for the fabric sample in which the simulation was applied. This is done using 20 methods. This allows for a comparison of the pre- and post-simulation situations. It provides suitable multi-angle image sets. In one application, the image set obtained before the simulation only shows the fabric type. It is fed into an artificial intelligence model (601) trained to identify the knitting of the fabric. It determines whether it is woven or has a special structure, and this information is used in the following stages. 25 It is used to select the image processing steps to be applied. In one application, the image acquisition unit (400) before and after simulation. The captured image sets are processed using a background algorithm. The background algorithm, By suppressing the fixed visual characteristics resulting from the natural texture of the fabric, simulation 30 This allows for the differentiation of surface changes that occur afterwards. As a result of this process, surface protrusions caused by hair growth and the shadow effects created by these protrusions are more It becomes noticeable. In one application, the images obtained after the reference algorithm are compared with 35 predefined ones. Depending on the type of fabric, it undergoes different pre-treatment stages. Knitted fabrics, woven fabrics... 18 Image processing methods defined separately for fabrics and specially structured fabrics. These pre-treatment stages are applied. These pre-treatment stages remove surface protrusions and various hair formations. to more clearly distinguish shadow structures formed under different lighting conditions It aims to... In an application, images that have gone through these preprocessing steps are used for both decision-making and... It is transferred to the evaluation unit (700) which carries out the rating process. The evaluation unit (700) assesses the presence and severity of feathering formations and It produces the result as a numerical quality score. The generated score is communicated with the user. through at least one user interface structured in such a way that, within a specific range and 10 It is presented as system output at a defined sensitivity level. The scope of protection of the invention is set out in the attached claims, and examples are provided in this detailed explanation. It cannot be limited to those explained for the purpose of this explanation. A person who is an expert in the field should provide the above-mentioned information. In light of these considerations, similar applications can be developed without deviating from the main theme of the invention. It is clear that he can place it.

Claims

19 REQUESTS 1. A system (100) for evaluating changes in the surface properties of fabrics, feature; - It will hold at least one fabric sample and the fabric sample will be within a 360-degree radius. a structured in such a way as to allow it to be moved to different angular positions platform (200); - configured to apply light to a fabric sample from multiple directions and angles lighting unit (300); - at least one simulation before and after, at various angular positions and with various lighting conditions. to record images of the fabric sample under the given conditions and multi-angle image sets an image acquisition unit configured to create (400); to process image sets with at least one reference algorithm and depending on the specified fabric type use an AI model structured to apply at least one preprocessing step. an image processing unit configured to process image sets (600), - using images that go through the processing stages to analyze fabric surface changes. a structured value for rating and creating a numerical quality score. licence unit (700); - The image acquisition unit (400) in question can write data to store it and the word the subject image processing unit (600), and the said evaluation unit (700) by data For storing non-volatile, computer-structured data that is readable, ready to be read. a memory readable by the wound (900); and - the platform in question (200), the lighting unit (300) and the image acquisition unit (400) a device (800) with a body designed to withstand (801); - located on the fuselage (801) and access to the components inside the fuselage (801) It includes an access element (802) that provides and - at least one simulation process of the system (100) carried out through a simulation unit before and after, textile fabric samples (200) found on the said platform The surface changes that occur are obtained by the multi-angle image acquisition unit (400). by analyzing the images through the image processing unit (600) and the artificial intelligence model (601) By determining the sample fabric types, the evaluation unit (700) evaluates them. in a way that will make It is structured.

2. According to claim 1, the system (100) has the feature that the artificial intelligence model in question is the fabric type. 35 It is structured in a way that determines whether it is knitted, woven, or specially made.

3. System (100) according to claim 1 or 2, and its feature is that the said artificial intelligence model, si- Trained to identify fabric type using a pre-implantation image set, and the fabric's to suppress the fixed visual characteristics resulting from its natural texture, pre-simulation and to process the post-simulation image sets with at least one of the aforementioned reference algorithms and structured to apply pre-treatment steps depending on the specified fabric type. It is.

4. System (100) according to any of the above requests, and its feature is; the reference in question The algorithm's ability to withstand changes in the fabric surface after the simulation process is problematic. to isolate the surface protrusions that emerge and the shadow effects created by these protrusions The problem is that the material is structured in a way that suppresses the natural texture of the metal.

5. According to any of the above requirements, the system is (100) and its characteristic is The image acquisition unit (400) has at least one of the following fabric types: knitted, woven and specially constructed fabrics. It is structured to execute processing steps.

6. System (100) according to any of the above requirements, and its feature is; image acquisition. unit (400), laser scanners configured to scan the fabric surface, projection data from at least one of the following: base-based light sources or lidar-based scanning systems. It is structured to collect data related to surface topography. 20 7. System (100) according to any of the above requirements, and its feature is; device (800) a mirror (803) positioned at a first angle on the base of the inside of its body (801) including; the plane on which the platform (200) extends, which allows the device body (802) to open outwards It includes an access element (802) that extends in a vertical plane; the aforementioned access element is 25 When a fabric is hung on (802), the access element (802) will make a second angle with it. It includes a mechanism (805) that enables its positioning; the first angle mentioned is image taking the unit has been chosen in such a way as to allow the mirror to see the entire fabric; The evaluation unit (700) will grade the wrinkle based on the image received. It is configured in this way. 30 According to requirement 7, a system (100) has the feature of having at least one wrinkle detection light unit. (804) access element (802) with the second angle mentioned on the surface remaining inside the body (801) structured to shine light from above onto the positioned fabric sample That is. 35 21 9. Occurrings in textile fabric samples before and after the simulation process. It is a method for evaluating surface changes, and its characteristic feature is that it includes the following steps: sidir: Obtaining a sample before simulation; a fabric sample at different angular positions and under different lighting conditions within a 360-degree range. Obtaining multi-angle image sets with their directions and angles; By analyzing the pre-simulation image set with a trained artificial intelligence model, the fabric type can be determined. to determine; Applying a simulation process to a fabric sample; Simulated at the same angular positions and under the same lighting conditions as before the simulation. To obtain multi-angle image sets of the fabric sample after simulation; Pre-simulation and post-simulation image sets, reflecting the natural texture of the fabric. a reference algorithm structured to suppress transmitted static visual features to do; Depending on the specified fabric type, pre-processing is applied to the images processed with the reference algorithm. to follow the steps; Presence of fabric surface defects on images that have undergone pre-processing stages. and by making a decision regarding its severity, to create a numerical quality score; and Presenting this numerical quality score in a user interface.

10. The method according to claim 9, and its feature is; the fabric construction of the artificial intelligence model (601). It is structured to determine probabilistically.

Images