Yarn weak twist and hairiness intelligent detection system based on visual stroboscopic synchronization

Abstract

The invention discloses a yarn weak twist and hairiness intelligent detection system based on visual stroboscopic synchronization, particularly relates to the field of quality control in the textile industry, is used for solving the problem of accurate detection of weak twist and hairiness defects in the yarn production process, and dynamically adjusts the flash frequency of a stroboscope according to the real-time speed and acceleration data of yarn, so that the detection accuracy is improved. The light source and the yarn are ensured to move synchronously, and a high-resolution image is obtained. In combination with yarn motion spectrum analysis and image quality evaluation, a complexity score and a matching degree score are generated and are used for dynamically regulating and controlling image noise reduction and edge detection parameters, and characteristic extraction requirements under different yarn motion conditions are met. The weak twist characteristic and the hairiness texture of the yarn are accurately extracted through a multi-scale edge detection and wavelet transform algorithm, a twist distribution model is constructed, the boundary range is corrected in combination with hairiness distribution data, characteristic interference is reduced, and the detection precision and reliability are improved.

Description

Technical Field

[0001] The present invention relates to the field of textile industry quality control, and more particularly to a yarn weak twist and hairiness intelligent detection system based on visual stroboscopic synchronization. Background Art

[0002] In the textile industry, accurate detection of yarn quality is an important part of ensuring textile production efficiency and the quality of finished products. The degree of weak twist and hairiness of yarn are two key indicators that affect yarn quality. Weakly twisted yarns are prone to breakage or uneven tension during weaving, while the presence of hairiness will directly affect the surface finish of the fabric and its subsequent processing performance. Traditional yarn quality detection methods mostly rely on manual observation or static analysis of a single indicator. Not only is the detection efficiency low, but the test results are easily affected by subjective factors, making it difficult to meet the needs of modern textile production for efficient, accurate, and real-time detection. With the development of production technology, automated detection equipment has gradually been applied to yarn detection. However, due to the complex and changeable movement state of the yarn, combined with the fixed parameter mode of traditional equipment, it is difficult to simultaneously meet the efficient and intelligent requirements of both weak twist and hairiness detection indicators.

[0003] The main problem currently lies in the complex motion characteristics of yarns during high-speed motion, such as vibration, jitter, and acceleration changes. Therefore, detecting weak twist areas and hairiness distribution often requires extremely high image clarity and feature resolution. However, in actual detection, due to inaccurate matching of the equipment's stroboscopic parameters with the yarn's motion state, image capture is prone to blurring, loss of detail, or noise interference. This results in unclear boundaries of twist variation areas during weak twist detection, making accurate feature extraction difficult. Furthermore, hairiness detection is subject to large measurement errors in hairiness length and density distribution due to variations in background light around the yarn and the yarn's inherent characteristics. Therefore, building an intelligent detection system that can adjust stroboscopic parameters in real time, optimize image quality, and achieve coordinated processing of weak twist and hairiness detection data has become a pressing technical challenge.

[0004] In order to solve the above problems, a technical solution is now provided. Summary of the Invention

[0005] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides an intelligent detection system for yarn weak twist and hairiness based on visual stroboscopic synchronization. First, according to the real-time speed and acceleration data of the yarn, the flash frequency of the stroboscope is accurately adjusted to ensure the high synchronization of the light source and the yarn movement, thereby obtaining a clear and detailed image. Subsequently, the spectrum analysis of the yarn motion characteristics and the image texture quality assessment are combined to generate a comprehensive complexity score and matching score. These scores are used to dynamically guide the noise reduction and edge detection parameter settings in the image processing process to ensure that key features can still be effectively extracted under different yarn motion conditions. Through multi-scale edge detection and wavelet transform algorithms, the weak twist characteristics and hairiness texture of the yarn can be accurately identified and separated, and a twist distribution model can be further constructed. The boundary correction is performed in combination with the hairiness distribution data, which significantly reduces the interference between characteristics and improves the accuracy and reliability of detection to solve the problems raised in the above-mentioned background technology.

[0006] To achieve the above object, the present invention provides the following technical solutions: A yarn weak twist and hairiness intelligent detection system based on visual stroboscopic synchronization includes: an image acquisition module, an image processing module, a feature extraction module, a weak twist detection module and a hairiness detection module.

[0007] Image acquisition module: The yarn speed sensor acquires real-time speed and acceleration data, and the time series fitting algorithm is used to calculate the stroboscope flash frequency to ensure that the flash is synchronized with the yarn movement. The industrial camera array is used to capture high-resolution images of the yarn from multiple angles, and the focus and lighting uniformity are corrected by combining blur feedback. The collected images and yarn motion data are then transmitted to the image processing module.

[0008] Image processing module: Combined with spectral analysis of yarn motion data, it extracts the trajectory fluctuation index, and simultaneously evaluates the edge clarity and noise distribution of image quality to generate a yarn complexity score and an image matching score. The scoring results are fed back to the image acquisition module to adjust the stroboscope frequency and camera acquisition parameters, and are also passed to the feature extraction module to guide the sensitivity settings of the noise reduction and feature extraction algorithms.

[0009] Feature extraction module: Based on the scoring results provided by the image processing module, the intensity of the wavelet denoising algorithm and the edge detection parameters are adjusted, and the weak twist characteristics and hairiness texture characteristics of the yarn are extracted through multi-scale edge detection; the extracted weak twist and hairiness data are transmitted to the weak twist detection module and the hairiness detection module respectively, and the feature extraction results are fed back to the image processing module.

[0010] Weak twist detection module: Receives weak twist characteristic data transmitted by the feature extraction module, establishes a twist distribution model through a segmented fitting method, and uses a gradient edge segmentation algorithm to identify the boundaries of weak twist regions. It then uses the hairiness distribution data provided by the hairiness detection module to correct the boundary range of the twist characteristics. The weak twist detection results are not only fed back to the image processing module to update the scoring model, but are also passed to the hairiness detection module as reference data for optimizing hairiness region segmentation.

[0011] Hairiness detection module: Based on the hairiness texture characteristics output by the feature extraction module, the morphological analysis algorithm is used to calculate the hairiness length and density distribution, and the region growing algorithm is used to segment the hairiness area. The hairiness segmentation range is corrected by combining the regional boundary characteristic data fed back by the weak twist detection module, and the hairiness detection results are passed to the image processing module to improve the scoring data.

[0012] In a preferred embodiment, the operation process of the image acquisition module includes the following: S1.1, the yarn speed sensor is used to collect the speed and acceleration data of the yarn in real time, and the instantaneous changes at each time interval are recorded to form a time series data set; the speed series is differentially analyzed to calculate its rate of change and extract the fluctuation amplitude in the short term; to characterize the frequency characteristics of the motion, the acceleration data is fast Fourier transformed to extract the main frequency component, which represents the main fluctuation frequency of the yarn motion.

[0013] S1.2 calculates the reference frequency of the yarn passing through the camera's field of view per unit time based on the collected speed mean and yarn diameter; then, the main frequency component is used as a correction term, combined with the fluctuation amplitude, to calculate the final stroboscopic frequency.

[0014] S1.3 The image captured by each camera is tested for blur after real-time transmission, and the clarity of the image edge texture is analyzed using the Laplace gradient method. When the blur value is lower than the set threshold, the camera's focus is adjusted and the lens position is fine-tuned through the real-time focus module until the blur index reaches the target value.

[0015] The images collected by S1.4 are divided into several areas after brightness distribution analysis. The average brightness and variance of each area are calculated to detect whether the lighting distribution is uniform. When the brightness difference in certain areas exceeds the uniformity threshold, local adjustments are made to areas with excessively low or high brightness by adjusting the brightness output power and incident angle of the controllable light source.

[0016] In a preferred embodiment, the operation process of the image processing module includes the following: S2.1 Based on the fluctuation amplitude and the main frequency component, the trajectory fluctuation index is constructed using the exponential function , used to characterize the complexity of yarn motion: ; Among them, the exponential factor Adjust the influence ratio of main frequency on trajectory fluctuation.

[0017] S2.2 evaluates whether the image can accurately capture the yarn motion characteristics and extracts edge clarity and noise distribution information from the image. The edge clarity of the image is calculated by the average value of the gradient change, using the formula: in, For the image at pixel points The gray value of represents the pixel gradient change, is the total number of pixels; at the same time, the image noise distribution is calculated by the variance of the grayscale change in the local area, and the formula is: in, For the The grayscale variance of the region, is the number of image partitions; in order to integrate edge clarity and noise distribution, an image clarity score is generated : .

[0018] S2.3 Yarn complexity scoring The trajectory fluctuation index and image clarity score are combined; the complexity score is calculated through a nonlinear combination model, and the formula is: .

[0019] S2.4 Image matching score Used to evaluate the consistency between the image and the yarn motion; call the trajectory fluctuation index and image clarity score, and calculate the matching degree by geometric mean: The resulting complexity score and matching score are then passed to the image acquisition module and feature extraction module.

[0020] In a preferred embodiment, the operation process of the feature extraction module includes the following: S3.1 adjusts the intensity of the wavelet transform denoising algorithm based on the yarn complexity score provided by the image processing module. The adjustment logic is based on a comprehensive evaluation of the yarn motion complexity and the image noise level. Specifically, the denoising intensity Calculated using the following formula: .

[0021] S3.2 Based on the image matching score, the feature extraction module adjusts the scale parameters of the multi-scale edge detection algorithm. Determined by the following calculation: in, is the parameter that controls the slope of the function, is the median threshold of the matching score.

[0022] S3.3 After the noise reduction and edge detection parameters are adjusted, the feature extraction module applies wavelet transform and multi-scale edge detection algorithm respectively to extract the weak twist characteristics and hairiness texture characteristics of the yarn from the processed image. Specifically, the weak twist characteristics and hairiness texture characteristics are extracted by the following steps: Weak twist feature extraction: Wavelet transform is used to decompose the image, identify the twist variation area, and extract the quantitative index of twist non-uniformity through local statistical analysis.

[0023] Hairiness texture feature extraction: Multi-scale edge detection is applied to capture the hairiness details around the yarn, the visibility of the hairiness area is enhanced through morphological operations, and the distribution density and length distribution characteristics of the hairiness are calculated.

[0024] The extracted characteristic data, including weak twist characteristics and hairiness texture features, are transmitted to the weak twist detection module and the hairiness detection module respectively.

[0025] In a preferred embodiment, the operation process of the weak twist detection module includes the following: S4.1 describes the twist distribution of the yarn, using a segmented fitting method to convert the yarn length Divide into multiple equal-length segments ; In each segment, use the polynomial fitting function Describes twist variation, where Represents the relative position of the yarn in the corresponding segment; optimize the polynomial coefficients by the least squares method to ensure the twist distribution model Can accurately reflect actual twist changes.

[0026] S4.2 Based on the twist distribution model, the gradient edge segmentation algorithm is applied to identify the boundaries of the area with significant twist changes; the specific operation is to calculate the gradient of the twist distribution , and set the threshold To determine the boundary position; when the twist gradient of a certain section exceeds the threshold, it is marked as the boundary of the weak twist area.

[0027] S4.3 adjusts the position of the twist boundary by analyzing the density and position deviation of the hairiness distribution; specifically, a correction function is set , which fine-tunes the boundary position of the twist distribution model according to the distribution of dense hairiness areas.

[0028] In a preferred embodiment, the operation process of the hairiness detection module includes the following: S5.1 further processes the hairiness characteristic data using a morphological analysis algorithm to quantify the length distribution of the hairiness and density distribution Specific steps include: Morphological opening: Removes small noise points in the hairy areas to preserve the main hair structure characteristics.

[0029] Morphological closing operation: fills small gaps in the hair area to enhance the integrity of the hair.

[0030] Calculate the length distribution of the treated hairiness area is the cumulative distance of the region boundary pixels, and the formula is: in, and is the coordinate difference of adjacent boundary pixels, is the total number of boundary pixels, Indicates the index of the pixel point; hairiness density distribution It is calculated by the number and total length of hairiness in a unit area: ;in, is the area of ​​the detection area, is the total number of hairs in the detection area.

[0031] S5.2 Based on morphological analysis, the region growing algorithm is applied to segment the hairiness region. The initial seed point of the region growing is selected from the centroid of the hairiness-dense region, and the region is gradually expanded according to the grayscale value difference of the neighborhood. The growing condition is defined as: in, is the grayscale value of the extended pixel, is the gray value of the seed point, is the grayscale threshold, Indicates the pixel number of the neighborhood inspection; through region growing, the distribution range of the hairiness is extracted and the segmented hairiness area image is generated.

[0032] S5.3 The gradient of the hairiness density and twist boundary area in the segmented area Perform joint analysis to determine whether there is a boundary error; when the overlap between a certain hairiness area and the weak twist boundary exceeds the set threshold, adjust the hairiness segmentation range : in, is the correction scale factor.

[0033] The technical effects and advantages of the intelligent yarn weak twist and hairiness detection system based on visual stroboscopic synchronization of the present invention are as follows: The present invention achieves high-precision synchronous detection of yarn weak twist and hairiness by comprehensively utilizing real-time yarn motion data acquisition, high-resolution multi-angle image capture, dynamic image quality assessment, and advanced feature extraction technology. First, based on the real-time speed and acceleration data of the yarn, the flash frequency of the stroboscope is precisely adjusted to ensure a high degree of synchronization between the light source and the yarn motion, thereby obtaining a clear and detail-rich image. Subsequently, by combining the spectrum analysis of the yarn motion characteristics with the image texture quality assessment, a comprehensive complexity score and matching score are generated. These scores are used to dynamically guide the setting of noise reduction and edge detection parameters in the image processing process, ensuring that key features can still be effectively extracted under different yarn motion conditions. Through multi-scale edge detection and wavelet transform algorithms, the weak twist characteristics and hairiness texture of the yarn can be accurately identified and separated, and a twist distribution model can be further constructed. The boundary correction is performed in combination with the hairiness distribution data, which significantly reduces the interference between the characteristics and improves the accuracy and reliability of the detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 The present invention is a structural schematic diagram of a yarn weak twist and hairiness intelligent detection system based on visual stroboscopic synchronization. DETAILED DESCRIPTION

[0035] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0036] Example 1: Figure 1 The invention provides an intelligent detection system for yarn weak twist and hairiness based on visual stroboscopic synchronization, which comprises an image acquisition module, an image processing module, a feature extraction module, a weak twist detection module and a hairiness detection module.

[0037] Image acquisition module: The yarn speed sensor acquires real-time speed and acceleration data, and the time series fitting algorithm is used to calculate the stroboscope flash frequency to ensure that the flash is synchronized with the yarn movement. The industrial camera array is used to capture high-resolution images of the yarn from multiple angles, and the focus and lighting uniformity are corrected by combining blur feedback. The collected images and yarn motion data are then transmitted to the image processing module.

[0038] Image processing module: Combined with spectral analysis of yarn motion data, it extracts the trajectory fluctuation index, and simultaneously evaluates the edge clarity and noise distribution of image quality to generate a yarn complexity score and an image matching score. The scoring results are fed back to the image acquisition module to adjust the stroboscope frequency and camera acquisition parameters, and are also passed to the feature extraction module to guide the sensitivity settings of the noise reduction and feature extraction algorithms.

[0039] Feature extraction module: Based on the scoring results provided by the image processing module, the intensity of the wavelet denoising algorithm and the edge detection parameters are adjusted, and the weak twist characteristics and hairiness texture characteristics of the yarn are extracted through multi-scale edge detection; the extracted weak twist and hairiness data are passed to the weak twist detection module and the hairiness detection module respectively, and the feature extraction results are fed back to the image processing module to correct the complexity scoring model and optimize the adaptability of the scoring algorithm.

[0040] Weak twist detection module: Receives weak twist characteristic data transmitted by the feature extraction module, establishes a twist distribution model through a segmented fitting method, and uses a gradient edge segmentation algorithm to identify the boundaries of weak twist regions. It then uses the hairiness distribution data provided by the hairiness detection module to correct the boundary range of the twist characteristics. The weak twist detection results are not only fed back to the image processing module to update the scoring model, but are also passed to the hairiness detection module as reference data for optimizing hairiness region segmentation.

[0041] Hairiness detection module: Based on the hairiness texture characteristics output by the feature extraction module, the morphological analysis algorithm is used to calculate the hairiness length and density distribution, and the region growing algorithm is used to segment the hairiness area. The hairiness segmentation range is corrected by combining the regional boundary characteristic data fed back by the weak twist detection module, and the hairiness detection results are passed to the image processing module to improve the scoring data.

[0042] In the textile industry, detecting yarn soft twist and hairiness is a critical step that impacts weaving process efficiency and finished product quality. However, due to the frequent fluctuations in speed and acceleration during high-speed yarn movement, as well as the complexity of the yarn's surface structure, traditional detection systems often face problems such as desynchronization, image blur, and uneven illumination. This directly reduces the accuracy of detecting soft twist and hairiness characteristics, making it difficult to meet the high demands of actual production for capturing detailed images and ensuring consistent detection. Therefore, the image acquisition module's task is not only to capture high-resolution images of the yarn but also to synchronize with the yarn's motion characteristics in real time, adjust the parameters of the stroboscope and industrial camera through precise calculations, and optimize the lighting and focus to generate high-quality images for subsequent analysis modules.

[0043] The operation process of the image acquisition module includes the following: S1.1, collect yarn speed and acceleration data in real time through yarn speed sensor, and record each time interval By performing differential analysis on the speed series and calculating its rate of change, the fluctuation amplitude in the short term can be extracted. In order to characterize the frequency characteristics of the motion, the acceleration data is subjected to fast Fourier transform to extract the main frequency components. , which represents the main fluctuation frequency of the yarn motion. This main frequency is combined with the speed mean and fluctuation range, which are used to describe the periodicity and non-steady-state characteristics of yarn motion and serve as important inputs for stroboscope frequency calculation.

[0044] S1.2 Based on the collected speed average and yarn diameter , calculate the reference frequency of the yarn passing through the camera field of view per unit time: Then, the main frequency component is used as a correction term, combined with the fluctuation amplitude, to calculate the final stroboscopic frequency: This frequency calculation process takes into account the effects of both the average yarn speed and speed fluctuations, ensuring that the stroboscope flash frequency perfectly matches the actual yarn movement, preventing image blurring due to flash misalignment.

[0045] The S1.3 industrial camera works according to the output frequency of the stroboscope and simultaneously captures high-resolution images of the yarn according to a preset multi-angle layout (such as 0 degrees, 45 degrees, and 90 degrees). The image collected by each camera is subjected to blur detection after real-time transmission, and the clarity of the image edge texture is analyzed using the Laplace gradient method. When the blur value is lower than the set threshold, the camera focus is adjusted. Adjustment: Use the real-time focus module to fine-tune the lens position until the blur index reaches the target value to ensure that the image clarity meets the requirements of subsequent analysis.

[0046] The image collected by S1.4 is divided into several areas after brightness distribution analysis, and the average brightness and variance of each area are calculated to detect whether the light distribution is uniform. When the brightness difference of some areas exceeds the uniformity threshold, the brightness output power of the controllable light source is adjusted. and the angle of incidence , making local adjustments to areas with low or high brightness. For example, increasing the light source power in low-brightness areas and reducing the intensity in high-brightness areas by adjusting the incident angle of the light source, thereby achieving uniform brightness distribution across the entire image.

[0047] S1.5 After the above processing, the final high-resolution image is generated together with the main frequency component, fluctuation amplitude, and fuzziness index. and illumination uniformity parameters The data is then passed to the image processing module. This data provides necessary support for the complexity score in the image processing module. At the same time, the blur and uniformity parameters are returned to the camera and light source control modules to form a closed-loop feedback loop, continuously optimizing the image acquisition quality.

[0048] The image acquisition module uses a yarn speed sensor to acquire real-time speed and acceleration data, constructing a yarn motion fluctuation model. The module then calculates the stroboscope's flash frequency based on the yarn's structural characteristics, achieving precise synchronization between the light source and the yarn's motion. The industrial camera captures high-resolution images from multiple angles and uses blur analysis feedback to adjust the focus in real time to ensure clear imaging. Furthermore, based on image brightness distribution analysis, the module adjusts the light source's output power and incident angle to improve illumination uniformity. The captured images and motion parameters not only provide foundational data for subsequent analysis but also optimize image quality through closed-loop feedback, ensuring the stability and efficiency of the inspection system.

[0049] During high-speed yarn motion detection, the image acquisition module calculates the trajectory fluctuation index and captures the image. However, the dynamic fluctuations and complex structure of the yarn can cause blur or noise in the captured image, which directly affects the accuracy of subsequent feature extraction. Therefore, the image processing module is tasked with conducting an in-depth analysis of image quality based on the trajectory fluctuation index transmitted by the acquisition module. By evaluating edge clarity and noise distribution, it calculates a yarn complexity score and an image matching score. These scores are not only the core basis for subsequent feature extraction, but also provide feedback to the acquisition module to optimize imaging parameters, forming a closed-loop adaptive optimization mechanism that provides reliable support for the high-precision detection of the entire system.

[0050] The operation process of the image processing module includes the following: S2.1 Based on the fluctuation amplitude and the main frequency component, the trajectory fluctuation index is constructed using the exponential function , used to characterize the complexity of yarn motion: ; Among them, the exponential factor Adjusts the proportion of the dominant frequency's influence on trajectory fluctuation. The trajectory fluctuation index quantifies the amplitude characteristics of yarn motion through the rate of change of velocity. Combined with the dominant frequency, it emphasizes the regularity and periodicity of yarn motion, providing a comprehensive assessment of the impact of yarn motion on image acquisition.

[0051] S2.2 In order to evaluate whether the image can accurately capture the yarn motion characteristics, it is necessary to extract edge clarity and noise distribution information from the image. The edge clarity of the image is calculated by the average value of the gradient change, and the formula is: in, For the image at pixel points The gray value of represents the pixel gradient change, is the total number of pixels. This formula quantifies the continuity and strength of edges in an image. The higher the clarity, the more complete the expression of image details.

[0052] At the same time, the image noise distribution is calculated by the variance of the grayscale change in the local area, and the formula is: in, For the The grayscale variance of the region, is the number of image partitions. A lower noise distribution value indicates less image signal interference.

[0053] To combine edge sharpness and noise distribution, generate an image sharpness score : The image clarity score reflects the image's ability to capture yarn details by enhancing edge clarity and suppressing noise distribution, providing support for subsequent scoring.

[0054] S2.3 Yarn complexity scoring The trajectory fluctuation index and image clarity score are combined. The complexity score is calculated through a nonlinear combination model, and the formula is: The logarithmic function in the formula is used to amplify the nonlinear contribution of the clarity score to the complexity score while limiting the bias introduced by noise and fluctuation.

[0055] The yarn complexity score reflects the multidimensional influence of yarn motion and image performance. The higher the score, the greater the motion complexity and the difficulty of capturing, providing a basis for subsequent algorithm parameter adjustment.

[0056] S2.4 Image matching score Used to evaluate the consistency between the image and the yarn motion. Call the trajectory fluctuation index and image clarity score, and calculate the matching degree by geometric mean: .

[0057] The resulting complexity score and matching score are passed to two modules: The data is passed to the image acquisition module to adjust the stroboscope frequency and camera acquisition parameters. For example, if the matching degree is low, the stroboscope frequency can be increased or the camera focus settings can be optimized.

[0058] It is passed to the feature extraction module as the sensitivity control parameter of the feature extraction algorithm to dynamically optimize the accuracy of noise reduction and feature extraction.

[0059] The image processing module combines yarn motion data with image quality analysis to establish a comprehensive calculation system for trajectory fluctuation index, image clarity score, and matching score. This module fully utilizes the interaction between motion data and image characteristics to accurately quantify the impact of yarn motion on image capture. It also optimizes image acquisition quality through a scoring feedback mechanism, providing key parameter support for the feature extraction module, achieving closed-loop control from data analysis to image capture, and laying the foundation for the dynamic adaptability and accuracy of the entire detection system.

[0060] In the yarn quality inspection system, the image acquisition module successfully captures and transmits yarn motion data and high-resolution images. The image processing module further analyzes this data to generate a yarn complexity score and an image matching score. These scores reflect the dynamic characteristics of the yarn during motion and the quality of the image capture, providing a critical reference for subsequent feature extraction. However, the score data alone is insufficient for accurate detection of weak twist and hairiness. Further image feature extraction is required to analyze the specific structural characteristics of the yarn. Therefore, the feature extraction module is developed. Its task is to optimize image processing parameters based on the score results provided by the image processing module and extract the weak twist characteristics and hairiness texture features of the yarn. This process not only provides accurate input data for the weak twist and hairiness detection modules, but also optimizes the scoring model through a feedback mechanism, improving the detection accuracy and intelligence level of the entire system.

[0061] The operation process of the feature extraction module includes the following: S3.1 adjusts the intensity of the wavelet transform denoising algorithm based on the yarn complexity score provided by the image processing module. The adjustment logic is based on a comprehensive evaluation of the yarn motion complexity and the image noise level. Specifically, the denoising intensity Calculated using the following formula: in, The function is used to introduce the nonlinear characteristics of the complexity score into the noise reduction strength. The effect of the matching score on the denoising strength is enhanced. This ensures that when the yarn motion is complex and the image matching is good, the denoising effect is moderately enhanced to remove random noise in the image while preserving the yarn's detailed features.

[0062] S3.2 Based on the image matching score, the feature extraction module adjusts the scale parameters of the multi-scale edge detection algorithm to adapt to the requirements of different image clarity and yarn characteristics. Determined by the following calculation: in, is the parameter that controls the slope of the function, is the median threshold of the matching score. This formula uses a logistic function, so that when the matching score is high, the scale parameter is small, enhancing the ability to detect subtle edges; when the score is low, the scale parameter is increased to capture rougher edge features, ensuring that yarn characteristics can still be effectively extracted even when the image quality is poor.

[0063] S3.3 After the noise reduction and edge detection parameters are adjusted, the feature extraction module applies wavelet transform and multi-scale edge detection algorithm respectively to extract the weak twist characteristics and hairiness texture characteristics of the yarn from the processed image. The weak twist characteristics are extracted by identifying the local changes and irregularities of the yarn twist, while the hairiness texture characteristics are obtained by analyzing the distribution and length changes of the fine fibers around the yarn. Specifically, the weak twist characteristics and hairiness texture characteristics Extract by the following steps: Weak twist feature extraction: Wavelet transform is used to decompose the image, identify the twist variation area, and extract the quantitative index of twist non-uniformity through local statistical analysis.

[0064] Hairiness texture feature extraction: Multi-scale edge detection is applied to capture the hairiness details around the yarn, the visibility of the hairiness area is enhanced through morphological operations, and the distribution density and length distribution characteristics of the hairiness are calculated.

[0065] The extracted characteristic data (weak twist characteristics and hairiness texture characteristics) are respectively transmitted to the weak twist detection module and the hairiness detection module for further characteristic analysis and detection.

[0066] After S3.4 completes feature extraction, the feature extraction module transmits the extracted weak twist characteristics and hairiness texture features to the corresponding detection modules. Simultaneously, the feature extraction module correlates the extracted features with the complexity score provided by the image processing module. By comparing the actual performance of the extracted features with the expected score, the feature extraction module adjusts the weight parameters of the complexity scoring model to optimize the adaptability and accuracy of the scoring algorithm. This feedback mechanism ensures that the scoring model's ability to assess image quality under different yarn types and motion conditions will gradually improve as the system accumulates operational data, further enhancing the overall intelligence of the detection system.

[0067] The feature extraction module dynamically adjusts the parameters of the noise reduction and edge detection algorithms by combining the complexity and matching scores provided by the image processing module to accurately extract the yarn's weak twist characteristics and hairiness texture features. A nonlinear combination formula is used to calculate the noise reduction and edge detection parameters, ensuring the adaptability and accuracy of feature extraction under varying image quality and yarn motion complexity. Furthermore, by feeding the extraction results back to the image processing module, the feature extraction module implements adaptive correction of the scoring model, optimizing the adaptability of the scoring algorithm and ensuring efficient and stable performance in a changing detection environment.

[0068] In the textile production process, yarn quality inspection is crucial to ensuring fabric performance and appearance. Yarn's weak twist characteristics and hairiness distribution are key indicators for assessing its quality. The image acquisition module captures real-time yarn motion data and high-resolution images. The image processing module further analyzes this data to generate a yarn complexity score and an image matching score. These scores reflect the dynamic characteristics of the yarn motion and the quality of the image capture, providing key references for the feature extraction module. Based on this, the feature extraction module extracts the yarn's weak twist characteristics and hairiness texture features, providing accurate data support for the weak twist detection and hairiness detection modules. However, the complexity of the yarn's motion and the presence of hairiness can lead to inaccurate identification of the weak twist region boundary. Boundary correction must be performed by incorporating hairiness detection data to reduce hairiness interference with weak twist detection and ensure high accuracy and reliability of the detection results. Therefore, the weak twist detection module uses a segmented fitting method to establish a twist distribution model based on the weak twist characteristic data transmitted by the feature extraction module. Combined with the hairiness distribution data provided by the hairiness detection module, this model accurately corrects the twist feature boundary range and optimizes the segmentation accuracy of the weak twist region.

[0069] Example 2: The operation process of the weak twist detection module includes the following: The weak twist detection module first receives weak twist characteristic data from the feature extraction module. This data contains local variations in yarn twist and indicators of unevenness. The module then performs a preliminary statistical analysis on the weak twist characteristic data, calculating the basic trends in twist variation and the distribution of outliers. This analysis identifies the amplitude and frequency of twist variations and determines the primary pattern of twist fluctuations, providing basic data for subsequent segmented fitting.

[0070] S4.1 In order to accurately describe the twist distribution of the yarn, a segmented fitting method is used to divide the yarn length into Divide into multiple equal-length segments In each segment, a polynomial fitting function is used Describes twist variation, where Represents the relative position of the yarn in the segment. The polynomial coefficients are optimized by the least squares method to ensure the twist distribution model This model accurately reflects actual twist changes. It integrates the twist change characteristics of each segment to form a detailed description of the overall twist distribution, providing a basis for identifying the boundaries of weak twist areas.

[0071] S4.2 is based on the twist distribution model. The module uses the gradient edge segmentation algorithm to identify the boundaries of areas with significant twist changes. The specific operation is to calculate the gradient of the twist distribution. , and set the threshold To determine the boundary location. When the twist gradient of a certain section exceeds the threshold, it is marked as the boundary of the weak twist area. Through this method, the module can accurately locate the starting and ending positions of the weak twist area, ensuring accurate identification of abnormal twist areas.

[0072] S4.3 To reduce the interference of hairiness on weak twist detection, the weak twist detection module receives the hairiness distribution data. By analyzing the density and position deviation of the hairiness distribution, an optimization algorithm is applied to adjust the position of the twist boundary. Specifically, a correction function is set , which fine-tunes the boundary position of the twist distribution model according to the distribution of the hairiness-dense area. For example, if the hairiness density near a certain twist boundary is Exceeding the preset threshold , then the twist boundary position is adjusted by the following formula : in, is the adjustment amplitude factor, represents the local variation amplitude of the twist distribution model, represents the first Segment yarn length segment, Indicates the first Through this correction, the module can dynamically adjust the twist boundary, reduce the interference of hairiness distribution on the identification of weak twist areas, and improve the accuracy of detection results.

[0073] After completing the identification and correction of the weak twist area, the weak twist detection module will output the final weak twist detection results. The data is then passed to the image processing module for updating the complexity scoring model. Simultaneously, the weak twist detection results are passed as reference data to the hairiness detection module, which optimizes the hairiness region segmentation algorithm. Through this feedback mechanism, the weak twist detection module not only performs high-precision twist detection itself but also provides data support for the optimization of other modules in the system, enhancing the intelligence and collaborative capabilities of the overall detection process.

[0074] The final weak twist detection results are output in the form of structured data, including the following: Region list (region boundary characteristic data): Each weak twist region contains the starting position, end position, mean twist, twist fluctuation, and hairiness interference correction amount.

[0075] Statistical report: including overall statistical indicators such as the total number of weak twist areas, average twist fluctuation range, average value and standard deviation of boundary clarity within the area.

[0076] Graphical distribution: Displays the spatial distribution and boundary positions of weak twist areas in the form of a twist distribution model curve, marking the areas affected by hairiness and the correction results.

[0077] The weak twist detection module uses a segmented fitting method to establish a yarn twist distribution model and, combined with a gradient edge segmentation algorithm, accurately identifies the boundaries of weak twist regions. A correction function dynamically adjusts the twist boundary position to reduce the interference of hairiness on twist detection, ensuring high accuracy and reliability of detection results. Furthermore, by feeding detection results back to the image processing and hairiness detection modules, the module enables adaptive updates of the scoring model and efficient linkage between system modules. This design not only improves the accuracy of weak twist detection but also enhances the system's robustness and intelligence in complex detection environments, ensuring the efficiency and stability of the overall detection process.

[0078] The hairiness distribution of yarns directly affects the surface quality and processing performance of fabrics. The feature extraction module performs a preliminary analysis of the hairiness texture around the yarn, generating characteristic data on hairiness length and density. These data lay the foundation for further distribution feature extraction and region segmentation. However, the complexity of the hairiness region and the irregularity of its boundaries make it difficult to accurately describe its true distribution range using a single morphological analysis. Furthermore, the hairiness region may overlap with the weak twist region, potentially leading to confusion in the identification of their characteristics. Therefore, the hairiness detection module not only accurately segments the hairiness region based on morphological analysis and a region growing algorithm, but also dynamically adjusts the segmentation range based on twist boundary data fed back by the weak twist detection module. This process not only enables the hairiness detection module to output highly accurate hairiness length and density distribution results, but also provides reliable data support for improving the scoring model and verifying the consistency of system detection.

[0079] Example 3: The operation process of the hairiness detection module includes the following contents: The hairiness detection module first receives the hairiness texture characteristic data from the feature extraction module. This data includes the length and density distribution of hairs around the yarn, as well as images of hairiness edge characteristics. The module analyzes the hairiness texture characteristic data and extracts preliminary information on the hairiness distribution, including the overall length range, local density, and edge complexity. This data serves as the input for subsequent morphological analysis and region segmentation.

[0080] S5.1 further processes the hairiness characteristic data using a morphological analysis algorithm to quantify the length distribution of the hairiness and density distribution The specific steps include: Morphological opening: Removes small noise points in the hairy areas to preserve the main hair structure characteristics.

[0081] Morphological closing operation: fills small gaps in the hair area to enhance the integrity of the hair.

[0082] Calculate the length distribution of the treated hairiness area is the cumulative distance of the region boundary pixels, and the formula is: in, and is the coordinate difference of adjacent boundary pixels, is the total number of boundary pixels, Indicates the index of a pixel or hair, used to accumulate length or other characteristics.

[0083] Hairiness density distribution It is calculated by the number and total length of hairiness in a unit area: in, is the area of ​​the detection area, is the total number of hairs in the detection area.

[0084] S5.2 Based on the morphological analysis, the region growing algorithm is applied to segment the hairy region. The initial seed point for region growing is selected from the centroid of the hairy dense region, and the region is gradually expanded based on the grayscale value difference of the neighboring regions. The growing condition is defined as: in, is the grayscale value of the extended pixel, is the gray value of the seed point, is the grayscale threshold, Indicates the pixel number of the neighborhood check or the hairiness region number, used for conditional judgment or region correction. Through region growing, the module can accurately extract the distribution range of hairiness and generate a segmented hairiness region image.

[0085] S5.3 The gradient of the hairiness density and twist boundary area in the segmented area Perform joint analysis to determine whether there is a boundary error. When the overlap between a certain hairiness area and the weak twist boundary exceeds the set threshold, adjust the hairiness segmentation range. : in, The adjustment result ensures the independence of the hairiness region boundary and the twist characteristic region, reduces boundary overlap interference, and improves segmentation accuracy.

[0086] The gradient of the twist boundary region is obtained by calculating the rate of change of the twist distribution model to quantify the variation characteristics of the yarn twist along the length direction. The twist distribution model is constructed based on a piecewise fitting polynomial function, and its gradient is calculated using the first-order derivative formula to represent the variation amplitude of the twist in the local area. In actual calculations, due to the discreteness of the data, the gradient value is approximated by the difference method, and the formula is ,in and is the length of the adjacent position, and By setting the gradient threshold, the position where the gradient value exceeds the threshold is marked as the twist boundary for subsequent characteristic analysis and regional correction.

[0087] The final hairiness test results include: Length and density distribution of hairiness.

[0088] Segmentation boundary image of the hairiness area and the corrected distribution range.

[0089] Hairiness statistics for the entire yarn, including average length, density and areal distribution pattern.

[0090] The detection results are passed to the image processing module to improve the yarn complexity score and image matching score of the scoring model.

[0091] The hairiness detection module quantifies the length and density distribution of hairiness through morphological analysis and accurately segments hairiness regions using a region-growing algorithm. By incorporating regional boundary characteristic data fed back by the weak twist detection module, the module dynamically adjusts the hairiness segmentation range, reducing interference from weak twist regions on hairiness segmentation and ensuring high accuracy and stability of detection results. The detection results not only provide precise data support for the image processing module to refine the scoring model but also verify the system's detection consistency through the full-process verification module, significantly enhancing the overall intelligence and robustness of yarn quality detection.

[0092] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters in the formulas are set by technicians in this field according to actual conditions.

[0093] The above description is merely illustrative of certain exemplary embodiments of the present invention. It goes without saying that those skilled in the art will be able to modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and description are illustrative in nature and should not be construed as limiting the scope of protection of the claims.

[0094] It should be noted that, in this document, if there are relational terms such as first and second, etc., they are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "comprises", "includes" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device that includes a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article or device. In the absence of further limitations, an element defined by the phrase "comprises a ..." does not exclude the presence of other identical elements in the process, method, article or device that includes the element.

[0095] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

Claims

1. An intelligent yarn weak twist and hairiness detection system based on visual stroboscopic synchronization, characterized in that: include: Image acquisition module, image processing module, feature extraction module, weak twist detection module and hairiness detection module; Image acquisition module: The yarn speed sensor acquires real-time speed and acceleration data, and a time series fitting algorithm is used to calculate the stroboscope flash frequency to ensure that the flash is synchronized with the yarn movement. An industrial camera array is used to capture high-resolution images of the yarn from multiple angles. Blur feedback is used to correct the focus and illumination uniformity, and the acquired images and yarn motion data are transmitted to the image processing module. Image processing module: Combined with spectrum analysis of yarn motion data, it extracts trajectory fluctuation index, evaluates edge clarity and noise distribution of image quality, and generates yarn complexity score and image matching score; The scoring results are passed to the feature extraction module to guide the sensitivity setting of the noise reduction and feature extraction algorithms; Feature extraction module: Based on the scoring results provided by the image processing module, the intensity of the wavelet noise reduction algorithm and the edge detection parameters are adjusted, and the weak twist characteristics and hairiness texture characteristics of the yarn are extracted through multi-scale edge detection. The extracted weak twist and hairiness data are transmitted to the weak twist detection module and the hairiness detection module respectively. Weak twist detection module: Receives weak twist characteristic data transmitted by the feature extraction module, establishes a twist distribution model through a segmented fitting method, and uses a gradient edge segmentation algorithm to identify the boundaries of weak twist regions. It then uses the hairiness distribution data provided by the hairiness detection module to correct the boundary range of the twist characteristics. The weak twist detection results are then passed to the hairiness detection module as reference data for optimizing hairiness region segmentation. Hairiness detection module: Based on the hairiness texture characteristics output by the feature extraction module, the morphological analysis algorithm is used to calculate the hairiness length and density distribution, and the region growing algorithm is used to segment the hairiness area. The hairiness segmentation range is corrected by combining the regional boundary characteristic data fed back by the weak twist detection module.

2. The intelligent yarn weak twist and hairiness detection system based on visual stroboscopic synchronization according to claim 1 is characterized in that: The operation process of the image acquisition module includes the following: S1.1, using a yarn speed sensor to collect yarn speed and acceleration data in real time, recording the instantaneous changes at each time interval to form a time series data set; performing differential analysis on the speed series to calculate its rate of change and extract the short-term fluctuation amplitude; to characterize the frequency characteristics of the motion, performing a fast Fourier transform on the acceleration data to extract the main frequency component, representing the main fluctuation frequency of the yarn motion; S1.2 calculates the reference frequency of the yarn passing through the camera field of view per unit time based on the collected speed mean value and yarn diameter; Then, the main frequency component is used as a correction term and combined with the fluctuation amplitude to calculate the final stroboscopic frequency; S1.3 The image captured by each camera is subjected to blur detection after real-time transmission, and the clarity of the image edge texture is analyzed using the Laplace gradient method; When the blur value is lower than the set threshold, the camera focus is adjusted and the lens position is fine-tuned through the real-time focus module until the blur index reaches the target value; The images collected by S1.4 are divided into several areas after brightness distribution analysis. The average brightness and variance of each area are calculated to detect whether the lighting distribution is uniform. When the brightness difference in certain areas exceeds the uniformity threshold, local adjustments are made to areas with excessively low or high brightness by adjusting the brightness output power and incident angle of the controllable light source.

3. The intelligent yarn weak twist and hairiness detection system based on visual stroboscopic synchronization according to claim 2, characterized in that: The operation process of the image processing module includes the following: S2.1 Based on the fluctuation amplitude and the main frequency component, the trajectory fluctuation index is constructed using the exponential function , used to characterize the complexity of yarn motion: ; Among them, the exponential factor Adjust the influence ratio of main frequency on trajectory fluctuation; S2.2 evaluates whether the image can accurately capture the yarn motion characteristics and extracts edge clarity and noise distribution information from the image. The edge clarity of the image is calculated by the average value of the gradient change, using the formula: in, For the image at pixel points The gray value of represents the pixel gradient change, is the total number of pixels; at the same time, the image noise distribution is calculated by the variance of the grayscale change in the local area, and the formula is: in, For the The grayscale variance of the region, is the number of image partitions; in order to integrate edge clarity and noise distribution, an image clarity score is generated : ; S2.3 Yarn complexity scoring The trajectory fluctuation index and image clarity score are combined; the complexity score is calculated through a nonlinear combination model, and the formula is: ; S2.4 Image matching score Used to evaluate the consistency between the image and the yarn motion; call the trajectory fluctuation index and image clarity score, and calculate the matching degree by geometric mean: The resulting complexity score and matching score are then passed to the image acquisition module and feature extraction module.

4. The intelligent yarn weak twist and hairiness detection system based on visual stroboscopic synchronization according to claim 3 is characterized in that: The operation process of the feature extraction module includes the following: S3.1 adjusts the intensity of the wavelet transform denoising algorithm based on the yarn complexity score provided by the image processing module. The adjustment logic is based on a comprehensive evaluation of the yarn motion complexity and the image noise level. Specifically, the denoising intensity Calculated using the following formula: ; S3.2 Based on the image matching score, the feature extraction module adjusts the scale parameters of the multi-scale edge detection algorithm. Determined by the following calculation: in, is the parameter that controls the slope of the function, is the median threshold of the matching score; S3.3 After the noise reduction and edge detection parameters are adjusted, the feature extraction module applies wavelet transform and multi-scale edge detection algorithm respectively to extract the weak twist characteristics and hairiness texture characteristics of the yarn from the processed image. Specifically, the weak twist characteristics and hairiness texture characteristics are extracted by the following steps: Weak twist feature extraction: Use wavelet transform to decompose the image, identify the twist variation area, and extract the quantitative index of twist non-uniformity through local statistical analysis; Hairiness texture feature extraction: Multi-scale edge detection is applied to capture the hairiness details around the yarn, morphological operations are used to enhance the visibility of the hairiness area, and the distribution density and length distribution characteristics of the hairiness are calculated; The extracted characteristic data, including weak twist characteristics and hairiness texture features, are transmitted to the weak twist detection module and the hairiness detection module respectively.

5. The intelligent yarn weak twist and hairiness detection system based on visual stroboscopic synchronization according to claim 4 is characterized in that: The operation process of the weak twist detection module includes the following: S4.1 describes the twist distribution of the yarn, using a segmented fitting method to convert the yarn length Divide into multiple equal-length segments ; In each segment, use the polynomial fitting function Describes twist variation, where Represents the relative position of the yarn in the corresponding segment; optimize the polynomial coefficients by the least squares method to ensure the twist distribution model Can accurately reflect actual twist changes; S4.2 Based on the twist distribution model, the gradient edge segmentation algorithm is applied to identify the boundaries of the area with significant twist changes; the specific operation is to calculate the gradient of the twist distribution , and set the threshold To determine the boundary position; when the twist gradient of a certain section exceeds the threshold, it is marked as the boundary of the weak twist area; S4.3 adjusts the position of the twist boundary by analyzing the density and position deviation of the hairiness distribution; specifically, a correction function is set , which fine-tunes the boundary position of the twist distribution model according to the distribution of dense hairiness areas.

6. The intelligent yarn weak twist and hairiness detection system based on visual stroboscopic synchronization according to claim 5, characterized in that: The operation process of the hairiness detection module includes the following: S5.1 further processes the hairiness characteristic data using a morphological analysis algorithm to quantify the length distribution of the hairiness and density distribution ; The specific steps include: Morphological opening: removes small noise points in the hairy area to preserve the main hair structure characteristics; Morphological closing operation: fills small gaps in the hair area to enhance the integrity of the hair; Calculate the length distribution of the treated hairiness area is the cumulative distance of the region boundary pixels, and the formula is: in, and is the coordinate difference of adjacent boundary pixels, is the total number of boundary pixels, Indicates the index of the pixel point; hairiness density distribution It is calculated by the number and total length of hairiness in a unit area: ;in, is the area of ​​the detection area, is the total number of hairs in the detection area; S5.2 Based on morphological analysis, the region growing algorithm is applied to segment the hairiness region. The initial seed point of the region growing is selected from the centroid of the hairiness-dense region, and the region is gradually expanded according to the grayscale value difference of the neighborhood. The growing condition is defined as: in, is the grayscale value of the extended pixel, is the gray value of the seed point, is the grayscale threshold, Indicates the pixel number of the neighborhood inspection; extracts the distribution range of the hairiness through region growing, and generates the segmented hairiness region image; S5.3 The gradient of the hairiness density and twist boundary area in the segmented area Perform joint analysis to determine whether there is a boundary error; when the overlap between a certain hairiness area and the weak twist boundary exceeds the set threshold, adjust the hairiness segmentation range : in, is the correction scale factor.

Images