A cloth quality on-line detection method and device for textile production

Abstract

The present application belongs to the technical field of textile quality detection, and relates to a fabric quality online detection method and device for textile production. A design original drawing of a digital printing pattern is obtained, and the design original drawing contains color channel information and pixel resolution information. Detection light is emitted to the fabric and transmitted light signals are collected to generate a light transmission image. The warp yarn density and weft yarn density are measured according to the light transmission image. The ratio of the yarn density to the design original drawing resolution is calculated to generate a printing capacity index, and the printing capacity index is corrected according to the color channel information. Fabric surface images are collected from at least two different angles, a three-dimensional morphology is reconstructed, and a fabric defect parameter set is extracted. The fabric quality score is comprehensively calculated according to the printing capacity index, the light transmission uniformity and the fabric defect parameter set, and an unqualified signal is output when the fabric quality score is lower than a preset threshold. The present application can evaluate the matching degree of the fabric physical structure and the design accuracy, and improve the accuracy of the online detection of printed fabric.

Description

Technical Field

[0001] This invention belongs to the field of textile quality testing technology, specifically a method and device for online testing of fabric quality in textile production. Background Technology

[0002] In textile printing production, online fabric quality inspection systems typically identify geometric deviations and color differences in printed patterns by acquiring images of the fabric surface and comparing them with standard samples. Some inspection solutions also incorporate light transmission imaging technology to measure the warp and weft density of the fabric, used to assess the fabric's basic physical parameters. These methods have been applied in industrial settings and can detect common quality problems such as printing misalignment, missing prints, and color differences.

[0003] However, existing detection methods still have room for improvement. The clarity of printed patterns depends not only on the printing process itself but also on the degree of matching between the fabric yarn density and the design resolution. When the fabric yarn density is insufficient to support the design resolution, jagged blurring or wavy distortion will appear at the edges of the printed pattern, problems that are difficult to identify in conventional image comparison and inspection. Simultaneously, different colored inks have different diffusion characteristics on fabrics, resulting in different printing effects for different colored patterns at the same yarn density, and existing methods have not yet incorporated color characteristics into the printing capability assessment. Furthermore, fabric tension fluctuations in the production line can cause distortion in yarn density measurements, and existing methods lack dynamic calibration methods based on prior information. Therefore, establishing a mapping relationship between the physical structure of the fabric and design parameters is one of the directions that those skilled in the art continue to focus on. Summary of the Invention

[0004] To address the problems mentioned in the background art, the present invention provides an online fabric quality detection method and apparatus for textile production, thereby solving the problem of low matching degree between fabric yarn density and printing design resolution.

[0005] To achieve the above objectives, the present invention provides the following technical solution: an online fabric quality detection method for textile production, comprising the following steps: Obtain the original design artwork of the digital printing pattern, which includes information on each color channel and pixel resolution; Detection light is emitted onto the fabric and the transmitted light signal passing through the fabric is collected. A light transmission image of the fabric is generated based on the transmitted light signal. Based on the alternating bright and dark light spots corresponding to the warp and weft yarns in the transmitted light image, the number of yarns per unit length in the warp direction and the number of yarns per unit length in the weft direction of the fabric are measured respectively to determine the warp yarn density and the weft yarn density. The basic printing capacity index is calculated based on the warp yarn density, weft yarn density, and pixel resolution of the original design. The basic printing capacity index is then corrected based on the color channel information in the original design to obtain the final printing capacity index. At the same time, the light transmission uniformity of the transmitted image is calculated. The surface images of the fabric are acquired from at least two different perspectives. The three-dimensional morphology of the fabric surface is reconstructed from the surface images, and the set of fabric defect parameters is extracted from the three-dimensional morphology. The fabric quality score is calculated by comprehensively considering the final printing capacity index, light transmission uniformity, and fabric surface defect parameters. When the fabric quality score is lower than a preset threshold, an unqualified signal is output.

[0006] Optionally, generating a translucent image includes: The detection light is emitted onto the fabric surface in the form of parallel light, and the emission wavelength of the detection light is dynamically selected according to the fabric color. Transmitted light signals are collected on the side of the fabric facing away from the light source to generate a multispectral transmittance image sequence of the fabric. Transmission feature values ​​of each spectral channel in a multispectral transmission image sequence are extracted, and these transmission feature values ​​are used to calculate color correction coefficients.

[0007] Optionally, when measuring the warp yarn density and weft yarn density, the following steps are included: The warp and weft interweaving structure of the fabric is transformed into a one-dimensional image of light and dark stripes through a cylindrical lens, so that the warp or weft yarns are clearly imaged rather than the yarn being measured is blurred. The light spot corresponding to each yarn in a one-dimensional bright and dark stripe image is converted into a photoelectric pulse signal; The original warp yarn density and the original weft yarn density were obtained by counting and measuring the period of the photoelectric pulse signal. The fabric stretch rate is calculated by using the difference between the preset pixel period in the original design and the actual pixel period after printing. Based on the fabric stretch rate, the original warp yarn density and the original weft yarn density are dynamically tension calibrated, and the final warp yarn density and the final weft yarn density are output.

[0008] Optionally, when calculating the basic printing capacity index, the following are included: Calculate the ratio of the final warp yarn density to the horizontal resolution of the original design, and the ratio of the final weft yarn density to the vertical resolution of the original design, respectively. The smaller of the two ratios is taken as the basic printing capacity index.

[0009] Optionally, the basic printing capacity index is modified to obtain the final printing capacity index, including: The color correction coefficient is calculated based on the proportion of each color channel in the original design and the light transmission characteristic values ​​of each spectral channel in the multispectral light transmission image sequence. The basic printing capability index is dynamically corrected based on the color correction factor to obtain the final printing capability index. When the final printing capacity index is lower than the first threshold of the corresponding color, it is determined that the fabric yarn density is insufficient to support the pixel resolution of the original design, and an insufficient resolution warning signal is output.

[0010] Optionally, when calculating the uniformity of light transmission, the following are included: Calculate the ratio of the standard deviation to the mean of all pixel gray values ​​in the transmitted image; Calculate the difference between the number 1 and the ratio, and use it as the light transmission uniformity; When the light transmission uniformity is lower than the second threshold, an early warning signal for uneven yarn density distribution is output.

[0011] Optionally, when acquiring a surface image of the fabric, the following steps are included: At least two cameras are used to simultaneously capture images of the fabric from the front and side views, respectively. After registering the multi-view images, three-dimensional point cloud data of the fabric surface is constructed. The following parameters are extracted from the 3D point cloud data: fluff density, fluff length, number of spherical defects, area of ​​spherical defects, height of spherical defects, wrinkle depth, wrinkle density, hole area, surface roughness, and surface smoothness.

[0012] Optionally, when inspecting the quality of the printed fabric, the overall fabric quality score includes: Obtain the geometric matching degree of the print; Weighting coefficients were assigned to the final printing capacity index, light transmittance uniformity, fabric defect parameter set, and printing geometric matching degree, respectively. The weighted sum of each item after assigning weight coefficients is used as the fabric quality score.

[0013] Optionally, after outputting an invalid signal, the following steps are included: Monitor the confidence level of the fabric quality score; When the confidence level is lower than the third threshold, the yarn density measurement, light transmission uniformity calculation and surface image acquisition are re-executed on the area corresponding to the fabric quality score with a higher scanning accuracy than the current resolution.

[0014] An online fabric quality inspection device for textile production includes: The original design acquisition module is configured to acquire the original design of the digital printing pattern, wherein the original design includes information on each color channel and pixel resolution information. The light transmission imaging module is configured to emit detection light onto the fabric and collect the transmitted light signal that passes through the fabric, generate a multispectral light transmission image sequence of the fabric based on the transmitted light signal, and extract the light transmission feature values ​​of each spectral channel. The yarn density detection module is configured to measure the yarn density in the warp and weft directions respectively based on the alternating bright and dark spots corresponding to the warp and weft yarns in the transmitted light image, and calculate the fabric stretch rate by using the difference between the preset pixel period in the original design and the actual pixel period after printing. After dynamic tension calibration of the measurement results, the final warp yarn density and the final weft yarn density are output. The printing capacity calculation module is configured to calculate the basic printing capacity index based on the final warp yarn density, the final weft yarn density and the pixel resolution of the design original, and calculate the color correction coefficient based on the color channel information in the design original and the light transmission characteristic value of each spectral channel in the multispectral light transmission image sequence. The basic printing capacity index is dynamically corrected to obtain the final printing capacity index, and the light transmission uniformity of the light transmission image is calculated at the same time. The 3D detection module is configured to acquire surface images of the fabric from at least two different perspectives, reconstruct the three-dimensional morphology of the fabric surface based on the surface images, and extract a set of fabric defect parameters from the three-dimensional morphology. The quality assessment module is configured to calculate the fabric quality score based on the final printing capability index, light transmission uniformity, and fabric surface defect parameter set. When the fabric quality score is lower than a preset threshold, an unqualified signal is output.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention provides an online fabric quality inspection method and apparatus for textile production. Its core lies in establishing a mapping relationship between design resolution and yarn density. By acquiring the original design of a digitally printed pattern, the ratio of the design resolution to the measured yarn density is calculated to generate a printability index. This allows the inspection system to directly determine whether the fabric possesses the physical conditions to support the design precision, rather than relying solely on post-printing pattern comparison. A color correction coefficient is calculated based on the proportion of each color channel in the original design, dynamically compensating for the printability index and distinguishing the printability criteria for different color patterns on the same fabric. The fabric elongation rate is calculated using the difference between the preset pixel period in the original design and the actual pixel period after printing, dynamically calibrating the yarn density measurement results to reduce the impact of tension fluctuations on the measured values. These technical means work together to form a printability assessment framework based on the original design, helping to predict the degree of matching between the fabric and design parameters before or during printing, thus improving the targeting of online inspection. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the overall method flow of the present invention; Figure 2 This is a schematic diagram of the internal module flow of the overall device of the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] like Figures 1 to 2 As shown, this invention provides an online fabric quality inspection method for textile production, comprising the following steps: Obtain the original design artwork of the digital printing pattern, which includes information on each color channel and pixel resolution; Detection light is emitted onto the fabric and the transmitted light signal passing through the fabric is collected. A light transmission image of the fabric is generated based on the transmitted light signal. Based on the alternating bright and dark light spots corresponding to the warp and weft yarns in the transmitted light image, the number of yarns per unit length in the warp direction and the number of yarns per unit length in the weft direction of the fabric are measured respectively to determine the warp yarn density and the weft yarn density. The basic printing capacity index is calculated based on the warp yarn density, weft yarn density, and pixel resolution of the original design. The basic printing capacity index is then corrected based on the color channel information in the original design to obtain the final printing capacity index. At the same time, the light transmission uniformity of the transmitted image is calculated. The surface images of the fabric are acquired from at least two different perspectives. The three-dimensional morphology of the fabric surface is reconstructed from the surface images, and the set of fabric defect parameters is extracted from the three-dimensional morphology. The fabric quality score is calculated by comprehensively considering the final printing capacity index, light transmission uniformity, and fabric surface defect parameters. When the fabric quality score is lower than a preset threshold, an unqualified signal is output.

[0019] Specifically, perform the following steps.

[0020] First, the original digital printing pattern design corresponding to the current production batch is retrieved from the production management system. This original design is stored as an electronic file, containing a distribution map of each color channel and the number of pixels per inch, i.e., horizontal and vertical resolution.

[0021] Next, the light transmission imaging device is activated. This device includes a linear parallel light source positioned above the fabric and a line scan camera positioned below the fabric. The light source emits parallel light at a 45-degree angle to the fabric plane, dynamically adjusting the emitted wavelength according to the fabric's background color. For example, a shorter wavelength of blue light is used for light-colored fabrics, and a longer wavelength of red light is used for dark-colored fabrics to enhance the contrast of the transmitted light. The line scan camera synchronously acquires the light signal transmitted through the fabric, with the sampling frequency of each row matched to the fabric's movement speed, generating a light transmission image row by row. In the light transmission image, dark spots are formed at the intersections of warp and weft yarns due to yarn overlap, resulting in reduced light transmission, while bright spots are formed at the gaps between the yarns, resulting in increased light transmission. The entire image presents a grid-like texture with alternating light and dark areas.

[0022] Then, yarn density analysis is performed on the transmitted image. Along the warp direction, the number of dark spots appearing per unit length is counted to obtain the warp yarn density. Along the weft direction, the number of dark spots appearing per unit length is counted to obtain the weft yarn density. Simultaneously, the preset pixel period of the printed pattern in the original design is obtained, which is the distance between adjacent pixels along the warp and weft directions in the original design. This preset pixel period is compared with the pixel period of the actual printed pattern on the current fabric; the difference reflects the deformation rate of the fabric due to tension stretching on the production line. Based on this deformation rate, the originally measured warp and weft yarn densities are inversely compensated to eliminate the effect of tension, and the final warp and weft yarn densities are output. For example, if the deformation rate is 5% stretching, the original density is divided by 1.05 to obtain the true density.

[0023] Next, the printing capacity index is calculated. The ratio of the final warp yarn density to the horizontal resolution of the original design is recorded as the first ratio, and the ratio of the final weft yarn density to the vertical resolution of the original design is recorded as the second ratio. The smaller of the first and second ratios is taken as the basic printing capacity index, which reflects the relative capacity of yarn density to the design resolution. Then, the area proportion of each color channel in the original design is read. Due to the different pigment particle sizes and surface tensions of different colored inks such as black, cyan, magenta, and yellow, their diffusion levels differ under the same yarn density. A color correction coefficient between 0.8 and 1.2 is calculated based on the proportion of each color channel. For example, a larger correction coefficient is used when the proportion of dark areas is high, indicating that a higher printing capacity is required. The basic printing capacity index is multiplied by this color correction coefficient to obtain the final printing capacity index. At the same time, the light transmission uniformity of the transmitted image is calculated: the gray values ​​of all pixels in the transmitted image are statistically analyzed, the ratio of its standard deviation to the mean is calculated, and then 1 is subtracted from the ratio. The closer the result is to 1, the more uniform the yarn distribution.

[0024] Next, 3D inspection is performed. Two industrial cameras are positioned along the fabric's path. The first camera is vertically aligned with the fabric surface, while the second is tilted at a 30-degree angle to the fabric plane. Both cameras simultaneously capture images, obtaining the front and side views of the fabric at the same moment. Using binocular vision principles, the two images are stereo matched to calculate the disparity of each pixel, thereby reconstructing a 3D point cloud of the fabric surface. Fabric defect parameters are extracted from the point cloud, including: the height and density of pile protrusions, the number and size of spherical defects, the depth and distribution of wrinkles, the boundary area of ​​holes, and the surface roughness and smoothness relative to the reference plane. All parameters are compiled into a fabric defect parameter set.

[0025] Finally, the final printing capacity index, light transmission uniformity, and fabric defect parameter set are input into the quality scoring model. This model is a pre-trained weighted summation formula, where the printing capacity index has a weight of 0.5, light transmission uniformity has a weight of 0.2, and the sum of the weights of all parameters in the fabric defect parameter set is 0.3. A fabric quality score between 0 and 100 is calculated. This score is compared to a preset quality threshold; if it is below the threshold, a non-conforming signal is sent to the sorting system, marking the current fabric segment as a defective product.

[0026] When generating a translucent image, the following steps are included: The detection light is emitted onto the fabric surface in the form of parallel light, and the emission wavelength of the detection light is dynamically selected according to the fabric color. Transmitted light signals are collected on the side of the fabric facing away from the light source to generate a multispectral transmittance image sequence of the fabric. Transmission feature values ​​of each spectral channel in a multispectral transmission image sequence are extracted, and these transmission feature values ​​are used to calculate color correction coefficients.

[0027] Furthermore, the steps for generating the light-transmitting image described above specifically include the following sub-steps.

[0028] A linear parallel light source is installed above the fabric's running plane. The light source's emission direction forms a 45-degree angle with the fabric's normal direction, the light spot width covers the entire fabric surface, and the length direction of the light spot is perpendicular to the fabric's movement direction. The emission wavelength of the light source is dynamically selected based on the fabric's background color. The dynamic selection steps are as follows.

[0029] First, a background color measurement was performed before formal testing. A spectroradiometer was used to collect the diffuse reflectance spectrum of the fabric surface, measuring wavelengths from 380 nm to 780 nm with sampling intervals of 5 nm. The average reflectance of three characteristic bands was extracted from the measured reflectance spectrum curve: the average reflectance of the red band from 620 nm to 750 nm. The average reflectance in the green band from 520 nm to 570 nm The average reflectance in the blue band from 450 nm to 480 nm The calculation formulas are as follows: ; ; ; in wavelength Reflectance at that location. Comparison , , The wavelength value is used to determine the primary color of the fabric, with the wavelength corresponding to the maximum value being selected. If the primary color is red, a green light source with a wavelength of 532 nm is chosen as the detection light; if the primary color is green, a red light source with a wavelength of 635 nm is chosen; and if the primary color is blue, a yellow light source with a wavelength of 590 nm is chosen. This wavelength selection strategy is based on the principle of complementary colors, which enhances the absorption of the detection light by the fabric and improves the contrast of the transmitted light. After the wavelength is selected, the light source driving circuit outputs the corresponding current to ensure that the light-emitting element operates stably at that wavelength.

[0030] Next, a hyperspectral linear array camera is installed on the side of the fabric facing away from the light source. This camera consists of an entrance slit, a collimating lens, a beam splitter grating, a focusing lens, and an area array detector. The entrance slit is 30 micrometers wide and its length matches the fabric width, with the slit direction parallel to the fabric's width direction. Transmitted light passing through the fabric travels in a straight line to the entrance slit, is confined by the slit, and then enters the collimating lens. The collimating lens converts the incident light into a parallel beam, which is projected onto the beam splitter grating. The beam splitter grating has a line density of 300 lines per millimeter, enabling it to disperse the incident parallel light into fan-shaped spectral bands according to wavelength. These spectral bands are then focused by the focusing lens onto different rows of the area array detector. The area array detector has 512 rows, corresponding to 512 spectral channels, each with a wavelength width of approximately 0.8 nanometers; the detector has 2048 columns, corresponding to 2048 sampling points along the fabric's width direction.

[0031] The camera's timing is synchronized with the fabric's movement. A rotary encoder is installed along the fabric's path; each encoder pulse indicates a 0.1 mm movement of the fabric. The encoder pulse triggers the camera to perform one exposure and readout. Each exposure lasts 100 microseconds, acquiring one line of spatial data. The acquired light signal is converted from analog to digital to obtain a spectral vector. ,in For the current row index, For the first The grayscale values ​​of each spectral channel in that row. The spectral vectors collected from all rows are stacked in chronological order to form a three-dimensional data cube. ,in For the width direction corresponding to the column index, The row index corresponds to the direction of movement. For the first The center wavelength of each spectral channel.

[0032] Assume that after acquisition, a multispectral transmittance image sequence is obtained, containing Row image. For each spectral channel Its corresponding two-dimensional image is ,in , Extract the transmittance characteristics of this channel, including the channel mean and channel standard deviation. Channel mean The calculation formula is: ; Channel standard deviation The calculation formula is: ; Arrange the mean and standard deviation of each channel in order to form a vector of light transmission characteristics. : ; The vector has a length of 1024, and each component is a floating-point number. Transmittance eigenvalue vector. The data is stored in system memory for subsequent color correction factor calculation steps. At this point, the steps of generating the translucent image and extracting translucent feature values ​​are complete.

[0033] When measuring warp and weft yarn density, the following is included: The warp and weft interweaving structure of the fabric is transformed into a one-dimensional image of light and dark stripes through a cylindrical lens, so that the warp or weft yarns are clearly imaged rather than the yarn being measured is blurred. The light spot corresponding to each yarn in a one-dimensional bright and dark stripe image is converted into a photoelectric pulse signal; The original warp yarn density and the original weft yarn density were obtained by counting and measuring the period of the photoelectric pulse signal. The fabric stretch rate is calculated by using the difference between the preset pixel period in the original design and the actual pixel period after printing. Based on the fabric stretch rate, the original warp yarn density and the original weft yarn density are dynamically tension calibrated, and the final warp yarn density and the final weft yarn density are output.

[0034] Furthermore, the above steps for measuring yarn density specifically include the following sub-steps.

[0035] First, a cylindrical lens array is installed in the transmitted light imaging optical path, positioned between the fabric's back-to-light source side and the hyperspectral linear array camera. The cylindrical lens array consists of multiple parallel cylindrical lens units, each with its axial direction parallel to the warp direction of the fabric. After passing through the fabric, the transmitted light first passes through the cylindrical lens array. The converging effect of the cylindrical lens units occurs only in the plane perpendicular to the yarn direction; light rays along the yarn direction remain unaffected. When the transmitted light passes through the cylindrical lens array, the warp rays are converged into a narrow band, while the weft rays remain divergent. By adjusting the distance between the cylindrical lens array and the camera's focal plane, the image corresponding to the warp is made clear and sharp, while the image corresponding to the weft is blurred. Conversely, if it is necessary to measure the weft yarn density, the cylindrical lens array is rotated 90 degrees so that the lens axis is parallel to the weft direction, resulting in a clear weft image and a blurred warp image.

[0036] Secondly, a linear array photoelectric sensor is installed at the focal plane of the cylindrical lens array. This sensor consists of a row of closely spaced photodiodes, each with a width corresponding to the projected width of a single yarn on the fabric. When the warp yarn is clearly imaged, the transmitted light spot corresponding to each warp yarn is focused onto a single photodiode. The transmitted light spot corresponding to the gap between adjacent yarns falls onto the gap between the photodiodes, forming an alternating bright and dark striped image. The photodiode converts the received light signal into a photocurrent, the magnitude of which is proportional to the light intensity. The photocurrent is larger at the warp yarn location and smaller at the gap location. The photocurrent output by each photodiode is converted into a voltage signal by a transimpedance amplifier. The amplitude of the voltage signal fluctuates periodically over time, with the fluctuation period corresponding to the yarn spacing.

[0037] Then, the voltage signal is fed into a pulse shaping circuit. This circuit consists of a comparator and a Schmitt trigger. The comparator's reference voltage is set to the midpoint of the voltage signal's peak-to-peak value. When the input voltage is higher than the reference voltage, the comparator outputs a high level; when the input voltage is lower than the reference voltage, the comparator outputs a low level. The Schmitt trigger shapes the comparator output, eliminating noise-induced glitches and outputting a standard rectangular pulse sequence. The rising edge of each pulse in the sequence corresponds to the position of a yarn. The pulse sequence is then fed into the input port of a counter, which counts the pulses within a fixed time window. Let the time window length be... The unit is seconds, and the distance the fabric moves within this time window is... ,in The speed of fabric movement is expressed in millimeters per second. The number of pulses counted within the time window is denoted as... The original yarn density The calculation formula is: ; in The unit is yarn per millimeter. Repeat the above measurement process to obtain the original warp yarn density. and original weft yarn density .

[0038] Next, dynamic tension calibration is performed. A preset pixel period is obtained from the original design. Let the horizontal resolution of the original design be... Pixels per inch, vertical resolution is Pixels per inch. Therefore, the physical spacing between adjacent pixels in the radial direction in the original design is... Millimeters, the physical spacing between adjacent pixels in the latitudinal direction is Millimeters. After the fabric printing is completed, an image of the printed pattern is captured by a camera, and the actual pixel period of the printed pattern is measured. Multiple consecutive printing units arranged along the warp direction in the printed pattern are selected, and the total length of these printing units in the image is calculated. This total length is then divided by the number of printing units to obtain the actual warp pixel period. The unit is millimeters. Similarly, the actual latitudinal pixel period is obtained. Then the meridional elongation and latitudinal elongation The calculations are as follows: ; ; like This indicates that the warp is stretched, resulting in a lower measured yarn density; if This indicates that the warp direction is compressed, resulting in an overestimation of the yarn density measurement. The formula for calibrating the original yarn density is: ; ; in For the final warp yarn density, The final weft yarn density is calculated. The calibrated density value is then output to the printing capacity calculation module. In this step, the cylindrical lens enables selective imaging in the warp and weft directions, photoelectric conversion and pulse counting enable automatic measurement of yarn density, and dynamic tension calibration eliminates the influence of fabric deformation on the density measurement results on the production line.

[0039] When calculating the basic printing capacity index, the following are included: Calculate the ratio of the final warp yarn density to the horizontal resolution of the original design, and the ratio of the final weft yarn density to the vertical resolution of the original design, respectively. The smaller of the two ratios is taken as the basic printing capacity index.

[0040] Furthermore, the steps for calculating the printing capacity index specifically include the following sub-steps.

[0041] First, obtain the final warp yarn density. and final weft yarn density All units are roots per millimeter. Obtain the horizontal resolution of the original design. and vertical resolution All units are pixels per inch. To convert yarn density units to threads per inch, use the following formula: ; ; Calculate the meridional ratio separately Ratio of lateral direction to lateral direction : ; ; The smaller of the two ratios is taken as the basic printing capacity index. : ; This basic printing capacity index reflects the relative printing capacity of yarn density to the design resolution; the smaller the value, the weaker the printing capacity.

[0042] The final printing capacity index is obtained by revising the basic printing capacity index, including: The color correction coefficient is calculated based on the proportion of each color channel in the original design and the light transmission characteristic values ​​of each spectral channel in the multispectral light transmission image sequence. The basic printing capability index is dynamically corrected based on the color correction factor to obtain the final printing capability index. When the final printing capacity index is lower than the first threshold of the corresponding color, it is determined that the fabric yarn density is insufficient to support the pixel resolution of the original design, and an insufficient resolution warning signal is output.

[0043] Furthermore, the steps to revise the basic printing capacity index to obtain the final printing capacity index specifically include the following sub-steps.

[0044] First, obtain the basic printing capacity index. This index is calculated using the steps described above. Simultaneously, the area percentage of each color channel is read from the original design. The original design is stored in the CMYK color space, and the area percentage of cyan is recorded as... The area ratio of magenta is denoted as The proportion of yellow area is denoted as The proportion of black area is recorded as And satisfy .

[0045] Secondly, the transmission feature values ​​of each spectral channel in the multispectral transmission image sequence are obtained. From the transmission feature value vector... Extract the mean value of the spectral channel corresponding to cyan. and standard deviation The center wavelength range of the cyan channel is 470 nm to 490 nm. The mean value of the spectral channel corresponding to magenta is extracted. and standard deviation The center wavelength range is 540 nm to 560 nm. The mean value of the spectral channel corresponding to yellow was extracted. and standard deviation The center wavelength range is 570 nm to 590 nm. The mean value of the spectral channel corresponding to black is extracted. and standard deviation The center wavelength range is 400 nanometers to 450 nanometers.

[0046] For each color, calculate its light transmission attenuation factor. Taking cyan as an example, the light transmission attenuation factor is... The calculation formula is: ; This factor ranges from 0 to 1; a smaller value indicates more uneven light transmission in that color channel, meaning more severe ink diffusion. The same calculation applies. , , .

[0047] Then, the color correction coefficient is calculated based on the area ratio of each color and the light transmission attenuation factor. The calculation formula is: ; The correction factor is set to 1 when the light transmittance attenuation factor for all colors is 0.8. Therefore, the final color correction factor is... for: ; This coefficient is typically between 0.8 and 1.5. Multiplying the basic printing capability index by the color correction factor yields the final printing capability index. : ; Finally, a first threshold is set. This first threshold is dynamically set based on the distribution of different color regions in the original design. For each pixel in the original design, its color category is determined based on its color channel value. If the cyan channel value of that pixel is greater than 128, it is classified as a cyan region, and the first threshold for that region is set accordingly. The default value is 0.7. Similarly, the first threshold for the magenta region... The default value is 0.75, which is the first threshold value for the yellow area. The default value is 0.65, which is the first threshold value for the black area. The default value is 0.85. For mixed-color areas, the weighted average of the threshold values ​​of each component color is taken as the first threshold for that area. The entire original design is iterated through, and the final printing capability index is compared pixel-by-pixel. The first threshold corresponding to the pixel. If If the density is less than this threshold, the fabric yarn density corresponding to that pixel location is determined to be insufficient to support the pixel resolution of the original design, and a resolution deficiency warning signal is output. This warning signal includes the location coordinates and deviation degree of the defective pixel. The calculation formula is: ; in This is the first threshold for that pixel location. The greater the deviation, the more severe the resolution mismatch. Warning signals can be displayed in real-time on the user interface or triggered by an alarm device.

[0048] When calculating light transmittance uniformity, the following should be included: Calculate the ratio of the standard deviation to the mean of all pixel gray values ​​in the transmitted image; Calculate the difference between the number 1 and the ratio, and use it as the light transmission uniformity; When the light transmission uniformity is lower than the second threshold, an early warning signal for uneven yarn density distribution is output.

[0049] Furthermore, the steps for calculating the uniformity of light transmission specifically include the following sub-steps.

[0050] First, a transmissive image is acquired from the transmissive imaging step. This transmissive image is a single-channel grayscale image, denoted as... Its resolution is pixels, of which This represents the total number of pixels along the width of the image. This represents the row number along the height of the image. Each pixel in the image... grayscale value The grayscale value ranges from 0 to 255. A grayscale value of 0 indicates that the location is completely opaque, corresponding to the area with the densest yarn overlap; a grayscale value of 255 indicates that the location is completely transparent, corresponding to the area with the largest yarn gaps. Under normal circumstances, the amount of light transmitted at the intersection of warp and weft yarns is reduced due to yarn overlap, resulting in a lower grayscale value; the amount of light transmitted increases at the yarn gaps, resulting in a higher grayscale value.

[0051] Next, calculate the arithmetic mean of the grayscale values ​​of all pixels in the transmitted image. The calculation formula is: ; This average value reflects the overall light transmittance of the entire image. Different fabrics, due to variations in thickness and density, will exhibit significantly different average light transmittance values.

[0052] Then, calculate the standard deviation of all pixel grayscale values ​​in the transmitted image. The calculation formula is: ; Standard deviation This reflects the degree of dispersion of gray values ​​in a transmitted image relative to the average value. When the fabric yarn density distribution is uniform, the difference in gray values ​​at different locations in the transmitted image is small. The value is relatively small; when the yarn density distribution is uneven, denser yarns in some areas lead to reduced light transmission and a lower gray value, while sparser yarns in other areas lead to increased light transmission and a higher gray value, resulting in a greater difference in gray values. The value is relatively large.

[0053] Next, calculate the ratio of the standard deviation to the mean. This ratio is denoted as . The calculation formula is: ; ratio This eliminates the influence of differences in overall light transmittance among different fabrics on the dispersion evaluation. For example, thicker fabrics have lower overall light transmittance, and the average value... Smaller, standard deviation It is also relatively small, but the ratio of the two is... It can objectively reflect the relative degree of dispersion; thin fabrics have higher overall light transmittance. Larger It is also relatively large, but It can still accurately reflect the uniformity of yarn distribution.

[0054] Then, calculate the light transmission uniformity. Light transmission uniformity is denoted as... It is defined as the ratio of the number 1 to the value of 1. The difference is calculated using the following formula: ; Light transmission uniformity The value range is from 0 to 1. When all pixel grayscale values ​​are exactly equal, , , This indicates that the yarn distribution is completely uniform. The greater the difference in grayscale values, the more... The larger, The larger, The closer to 0, the more uneven the yarn distribution. Light transmission uniformity. It is a dimensionless pure numerical value, unaffected by the overall brightness of the image, and can be used for lateral comparison between fabrics of different materials and batches.

[0055] Light transmission uniformity The physical meaning can be further explained as follows: The closer it is to 1, the more uniform the yarn density distribution. When printing on the fabric, the absorption and diffusion of ink will tend to be uniform, and it will be less likely to produce patches of varying color depth. The lower the value, the more uneven the yarn density distribution, which may result in defects such as localized areas of excessively dark or light color during printing.

[0056] Next, set a second threshold. Second threshold The value is pre-calibrated based on the fabric type, printing precision requirements, and application scenario. For ordinary printed fabrics, such as home textiles and ordinary clothing fabrics, the second threshold... It can be set to 0.85. For high-precision printed fabrics, such as digital printing and high-resolution patterned fabrics, the second threshold... The threshold is set to 0.92. This threshold can be obtained by acquiring light transmission images of a batch of qualified fabrics, calculating their light transmission uniformity, taking the lowest light transmission uniformity value among all qualified samples as the initial threshold, and then fine-tuning it based on quality feedback in actual production.

[0057] The calculated light transmittance uniformity With the second threshold Comparison. When When the yarn density distribution is uneven, it indicates that the fabric's light transmittance uniformity is below the acceptable standard and the unevenness exceeds the allowable range. At this time, the system outputs a warning signal for uneven yarn density distribution. This warning signal contains the location information of the uneven area.

[0058] The specific method for obtaining the location information of uneven regions is as follows. A sliding window is defined, with a window size of [size missing]. pixels, of which The value is set to 32. The sliding window starts from the top left corner of the translucent image and moves in steps along both the horizontal and vertical directions. Movement, step size The value is set to 16 pixels. At each window location, a subset of the grayscale values ​​of all pixels within the window is extracted. Following the method described above for calculating transmittance uniformity, the local transmittance uniformity of that window is calculated. .when When the area covered by the window is determined to be a region with uneven yarn density distribution, the center pixel coordinates of the window are recorded. After traversing the entire image, all recorded window center coordinates are clustered according to connectivity, and duplicate or overlapping regions are removed. Finally, the bounding box coordinates of one or more non-uniform regions are output. Each bounding box is represented by the coordinates of its top-left and bottom-right corners, in pixels.

[0059] After the warning signal is output, one or more of the following actions can be triggered: highlighting the location of the uneven area on the operating interface; sending a signal to the alarm device to issue an audible and visual alarm; sending an instruction to the sorting system to mark the corresponding fabric segment as a defective product; and storing the warning information in the production database for process traceability and quality analysis. Simultaneously, this warning signal can form a closed loop with subsequent tension calibration or printing parameter adjustments. For example, when multiple consecutive batches show low light transmittance uniformity, it prompts operators to check the tension control or yarn supply in the weaving process.

[0060] In this step, the average value Standard deviation ,ratio Light transmission uniformity The step-by-step calculation ensures the objectivity and comparability of the light transmittance uniformity index; the sliding window positioning method enables the precise location and marking of uneven areas; and the dynamic setting of the second threshold adapts to the product requirements of different precision levels. Light transmittance uniformity, as an independent detection dimension, complements the printing capability index and the fabric defect parameter set, jointly participating in the comprehensive evaluation of fabric quality scores.

[0061] When acquiring surface images of a fabric, the following are included: At least two cameras are used to simultaneously capture images of the fabric from the front and side views, respectively. After registering the multi-view images, three-dimensional point cloud data of the fabric surface is constructed. The following parameters are extracted from the 3D point cloud data: fluff density, fluff length, number of spherical defects, area of ​​spherical defects, height of spherical defects, wrinkle depth, wrinkle density, hole area, surface roughness, and surface smoothness.

[0062] Furthermore, the steps of acquiring surface images of the fabric and extracting a set of fabric defect parameters specifically include the following sub-steps.

[0063] First, a multi-view image acquisition device was constructed. This device includes two industrial cameras, two linear light sources, and a set of plane mirrors. The first camera is mounted directly above the fabric, with its optical axis perpendicular to the fabric plane, a lens focal length of 25mm, and a field of view of 300mm. The second camera is mounted to the side of the fabric, with its optical axis at a 30-degree angle to the fabric plane, a lens focal length of 50mm, and a field of view of 200mm. Both cameras have a sensor resolution of 2448px × 2048px and a pixel size of 3.45μm. The two linear light sources are mounted on the same side of the two cameras, at a 30-degree angle to the camera optical axis, providing uniform illumination. The mirror set includes two plane mirrors, each positioned on the opposite side of the fabric at a 45-degree angle to the fabric plane, allowing the second camera to simultaneously capture the frontal reflected image and the side direct image of the fabric, thus obtaining images from three different perspectives: a front view, a left side view, and a right side view.

[0064] Secondly, two cameras are simultaneously triggered to acquire images. A rotary encoder is installed along the fabric's movement path; each encoder pulse signal represents a fabric movement of 0.1 mm. The encoder pulse signals simultaneously trigger the exposure of both cameras: the first camera's exposure time is 50 μs, and the second camera's is 80 μs. The acquired image data is sent to the image processing unit. The image acquired by the first camera is denoted as... The image captured by the second camera is recorded as ,in , These are pixel coordinates. Image segmentation algorithms are used to... The direct imaging region and the mirror region are separated from each other, and are denoted as follows: and This yields the front view, left side view, and right side view.

[0065] Then, the multi-view images are registered. First, distortion correction is performed on the three images to eliminate geometric distortion caused by lens distortion. The distortion correction model is as follows: ; ; in The coordinates of the principal point in the image. , , The lens distortion coefficients are obtained through camera calibration. Then, a scale-invariant feature transform algorithm is used to extract feature points from the three images. These feature points are then matched to establish pixel correspondences between different viewpoints. After matching, the depth value of each pixel is calculated based on the principles of binocular stereo vision. : ; in Baseline distance, which is the horizontal distance between the optical centers of the first and second cameras, in mm; The focal length of the camera is expressed in pixels (px). Let be the horizontal coordinates of the point in the left image. Let be the horizontal coordinate of the point in the right-hand image, and let be the difference between the horizontal coordinate and the horizontal coordinate, which is the disparity. After calculating the depth value for each pixel, a depth map is obtained. Depth map With front view Combined, a 3D point cloud data is generated. The coordinates of each point in the point cloud are... ,in , Obtained from image pixel coordinates based on camera intrinsic parameters. This is the depth value, in mm.

[0066] Next, fabric defect parameters are extracted from the 3D point cloud data. These parameters include five categories: nap, spherical defects, wrinkles, damage, and surface morphology. The total number of points in the point cloud is denoted as [missing information]. The physical area covered by the point cloud is , unit mm².

[0067] Extraction of pile parameters: A plane fitting algorithm is used to fit the reference plane of the fabric body. The equation of the reference plane is expressed as: ; in , , These are the fitting coefficients. Calculate the perpendicular distance from each point in the point cloud to the reference plane. .Will The points are identified as fluffy dots, and the number of fluffy dots is recorded as follows: Pile density The calculation formula is: ; Pile length Defined as the average distance of all fibers from the reference plane: ; Extraction of spherical defect parameters: The point cloud was clustered and segmented to identify clusters with a protrusion height greater than 0.5 mm and a planar projected area less than 4 mm². Each cluster corresponds to one spherical defect. The number of spherical defects... The value is directly given by the number of clusters. For each spherical defect, its projected area... The height of spherical defects is obtained by calculating the convex hull area of ​​the clustered point cloud on the reference plane, in mm². Defined as the maximum distance from each point within the cluster to the reference plane, in mm. The average area of ​​all spherical defects is denoted as... The average height is recorded as .

[0068] Extraction of fold parameters: Calculate the curvature variation of the point cloud along the meridional and latitudinal directions. Meridional curvature. The calculation formula is: ; in The distance between adjacent points along the longitudinal direction, in mm. Zonal curvature. The calculation formula is: ; in This represents the distance between adjacent points in the latitudinal direction, in mm. Points with an absolute curvature value greater than 0.05 are marked as fold points, and the number of fold points is denoted as [missing information]. . Fold density Defined as: ; fold depth Defined as the maximum height difference between the highest and lowest points in the local area where the fold point is located, in mm.

[0069] Extraction of damage parameters: Searching in point cloud A region with more than 50 points whose coordinates are more than 0.5 mm below the reference plane. Each such region corresponds to a hole. Hole area. The area is obtained by calculating the projected area of ​​the region on the reference plane, in mm².

[0070] Extraction of surface morphology parameters: surface roughness Defined as the average of the absolute values ​​of the distances from all point cloud points to the reference plane: ; Surface flatness Defined as the standard deviation of point cloud height values: ; in Average height of all point cloud points, in mm. Flatness. The smaller the value, the smoother the fabric surface.

[0071] Combine all the extracted parameters into a set of fabric defect parameters. : When inspecting the quality of printed fabric, the comprehensive fabric quality score includes: Obtain the geometric matching degree of the print; Weighting coefficients were assigned to the final printing capacity index, light transmittance uniformity, fabric defect parameter set, and printing geometric matching degree, respectively. The weighted sum of each item after assigning weight coefficients is used as the fabric quality score.

[0072] Furthermore, when inspecting the quality of printed fabric, the steps for comprehensively calculating the fabric quality score specifically include the following sub-steps.

[0073] First, obtain the printing geometric matching degree. The method for obtaining the printing geometric matching degree is as follows.

[0074] The original design image is read from the original design acquisition module. The original design image is a digital file of the digital printing pattern, denoted as... ,in , Here are the pixel coordinates. The printed image of the current fabric is obtained from the fabric surface image captured by the front-view camera, denoted as... .

[0075] Image of fabric surface With the original design image Perform image registration. The purpose of registration is to align two images to the same coordinate system for pixel-by-pixel comparison. The registration process includes the following sub-steps.

[0076] The first step is feature point extraction. An accelerated and robust feature extraction algorithm is used to extract feature points from the original design image, denoted as the set. ,in This is the feature description vector. The same algorithm is used to extract feature points from the fabric surface image, denoted as the set. .

[0077] The second step is feature point matching. For For each feature point in the dataset, calculate its relationship with... The Euclidean distance between the feature point description vectors is used. The feature points corresponding to the closest and second-closest distances are found. Let the closest distance be denoted as . The second closest distance is .like If the match is not found, then the matching pair is retained. After the above filtering, an initial set of matching point pairs is obtained. .

[0078] The third step is transformation model estimation. The fabric surface image may have undergone translation, rotation, and scaling relative to the original design image. An affine transformation model is used to describe the geometric mapping relationship between the two images. Affine transformation matrix. The form is: ; The corresponding coordinate transformation relationship is: ; A random sampling consensus algorithm is used to remove mismatched points from the initial set of matching point pairs and to estimate the optimal affine transformation matrix. The random sampling consensus algorithm is set to 1000 iterations, and the distance threshold is set to 3 pixels.

[0079] Step 4: Image resampling. This involves resampling the original design image. According to the affine transformation matrix Resampling is performed to generate the registered original design image. To make it match the image of the fabric surface Spatially aligned. Resampling uses bilinear interpolation.

[0080] Next, calculate the pixel difference between the two registered images. Since the brightness of the two images may differ, brightness adjustment is necessary first. Calculate the average grayscale value of each registered original design image. and the average gray value of the fabric surface image and the standard deviation of the registered original design image. Standard deviation of fabric surface image The grayscale values ​​of the fabric surface image are linearly transformed to match the mean and standard deviation of the original design image. The transformation formula is: ; Then, calculate the absolute value of the grayscale difference pixel by pixel. : ; The absolute value of the grayscale difference is greater than a preset threshold. Pixels that are not clearly distinguishable are marked as difference pixels. Preset threshold. The value is 30. The total number of differing pixels is denoted as... The total number of pixels in the image is denoted as... .

[0081] Finally, the geometric matching degree of the print is generated based on the pixel difference value. The formula for calculating the geometric matching degree of the print is: ; Print geometric matching degree The value range is from 0 to 1. When the two images are completely identical, , When the two images are completely different, , .

[0082] Secondly, obtain the final printing capacity index, denoted as... The light transmittance uniformity is obtained and denoted as... Obtain the set of fabric defect parameters, which includes the pile density. Length of fluff Number of spherical defects Average area of ​​spherical defects Average height of spherical defects fold density fold depth Area of ​​the hole Surface roughness and surface flatness .

[0083] Then, weighting coefficients are assigned to each of the above parameters. In this embodiment, the weight of the printing capacity index is... Set the weight of light transmittance uniformity to 0.35. Set to 0.15, the weight distribution of each parameter in the fabric defect parameter set is as follows: the sum of the weights of pile density and pile length is 0.10, where the weight of pile density is... Pile length weight The sum of the weights for the number of spherical defects, the average area of ​​spherical defects, and the average height of spherical defects is 0.20, where the weight for the number of spherical defects is... Weight of average area of ​​spherical defects Weight of average height of spherical defects The sum of the weights for fold density, fold depth, and hole area is 0.20, where fold density has the largest weight. Fold depth weight Hole area weight ; Weight of surface roughness Weight of surface smoothness The sum of all weights is 1.

[0084] Printing Capacity Index The value ranges from 0 to 1 and is directly included in the weighted summation. Light transmittance uniformity. The value ranges from 0 to 1, and it directly participates in the weighted summation.

[0085] For each parameter in the fabric defect parameter set, it is necessary to convert them into score values, so that the smaller the parameter value, the higher the score. Taking pile density as an example, the pile density measurements of 100 sets of fabric are collected in advance, and the maximum value is counted. and minimum value Current fabric pile density score. The calculation is as follows: ; A score of 1 is given when the pile density is at its minimum, and a score of 0 is given when it is at its maximum. Pile length score. The calculation method is the same: ; Score for the number of spherical defects for: ; Average area score of spherical defects for: ; Average height score of spherical defects for: ; Wrinkle density score for: ; Wrinkle depth score for: ; Hole area score for: ; Surface roughness score for: ; Surface flatness score for: ; Each of the above scores ranges from 0 to 1. Multiplying each score by its corresponding weight and summing the results yields the overall score for the set of fabric defect parameters. : ; Substituting the weights into the values: ; Then, the final printing capacity index Light transmission uniformity The overall score of the set of fabric defect parameters Geometric matching degree of print A weighted summation is performed. In this embodiment, the weight of the printing capacity index is adjusted to... The weight of light transmittance uniformity is adjusted to The weighting of the overall score for fabric defects has been adjusted to... The weight of the geometric matching degree of the print is adjusted to The sum is 1.

[0086] Fabric quality score The calculation formula is: ; Substituting the weights into the values: ; Calculation results Multiply by 100 to get a percentage score. Then, compare the percentage score with a preset threshold. Compare. Preset threshold. Product grades are divided into three levels: superior grade (threshold 90 points), first grade (threshold 75 points), and qualified grade (threshold 60 points). When the fabric falls below the threshold corresponding to the current product grade, a non-compliance signal is output, indicating that the current fabric segment does not meet the quality requirements. Simultaneously, scores for each sub-item are output for quality traceability.

[0087] After outputting an invalid signal, the following are included: Monitor the confidence level of the fabric quality score; When the confidence level is lower than the third threshold, the yarn density measurement, light transmission uniformity calculation and surface image acquisition are re-executed on the area corresponding to the fabric quality score with a higher scanning accuracy than the current resolution.

[0088] Furthermore, the steps following the output of the non-compliant signal specifically include the following sub-steps.

[0089] First, monitor the fabric quality score. The confidence level is used to measure the reliability of the current test results, and its calculation method is as follows. In the comprehensive calculation process described above, all input parameters have measurement uncertainties. The final printing capability index is calculated separately. Confidence level, light transmission uniformity Confidence level, comprehensive score of fabric defect parameter set Confidence level and printing geometric matching degree The confidence level of the fabric quality score is then calculated, and the overall confidence level of the fabric quality score is synthesized.

[0090] Final printing capacity index confidence level This depends on the repeatability of the yarn density measurement and the stability of the color correction coefficient. In the yarn density measurement step, 10 frames of transmitted light images are continuously acquired, and the final warp yarn density and final weft yarn density corresponding to each frame are calculated. Let the sequence of warp density values ​​obtained from the 10 measurements be... Calculate its mean and standard deviation The coefficient of variation for the meridional density measurement is: ; Similarly, the coefficient of variation for latitudinal density measurements is obtained. The larger of the two values ​​is taken as the coefficient of variation for yarn density measurement. The confidence level of the final printing capacity index is defined as follows: ; Light transmission uniformity confidence level This depends on the signal-to-noise ratio of the transmitted image. Calculate the average gray value of the signal region in the transmitted image. and the average gray value of the background area The signal region is the area covered by the yarn, and the background region is the area between the yarns. The signal-to-noise ratio is defined as: ; in The standard deviation of gray levels in textureless regions of an image can be obtained by photographing a uniform white board. The confidence level of transmittance uniformity is defined as: ; when hour, ;when hour, .

[0091] Overall score of fabric defect parameters confidence level This depends on the point cloud density and registration error of the 3D reconstruction. Let the total number of points in the 3D point cloud be... The physical area covered by the point cloud is The point cloud density is Unit: cloud density per square millimeter. (Based on standard point cloud density.) Points per square millimeter. The confidence level decreases proportionally when the actual point cloud density is lower than the standard density. Disparity dispersion in stereo matching is also considered. Disparity dispersion is defined as the standard deviation of all pixel disparity values. with mean disparity The ratio: ; The confidence score for the comprehensive score of fabric defect parameters is calculated using the following formula: ; Print geometric matching degree confidence level This depends on the number of valid matching point pairs and the registration residual in image registration. Let the number of valid matching point pairs after filtering by the random sampling consensus algorithm in the registration step be... The registration residual is defined as the average reprojection error of all valid matching point pairs after transformation. ; in This represents the coordinates of the feature points of the original design after affine transformation. The confidence score for the geometric matching degree of the print is calculated as follows: ; When the number of valid matching point pairs reaches 50 or more, this contribution is 1; when the reprojection error is less than 5 pixels, this contribution is close to 1.

[0092] Fabric quality score Overall confidence level The weighted average of the four confidence levels is taken, with each item having the same weight in the fabric quality score: ; Substitution , , , ,have to: ; Overall confidence level The value range is from 0 to 1.

[0093] Then, set the third threshold. The third threshold Based on production requirements, the value in this embodiment is set to 0.75. When If the confidence level of the current test result is deemed insufficient, a second scan is triggered.

[0094] After triggering a second scan, the system scores the fabric quality with a higher scanning accuracy than the current resolution. For the corresponding defective areas, yarn density measurement, light transmittance uniformity calculation, and surface image acquisition are performed again. Specifically, the location coordinates of the fabric area corresponding to the defective signal are obtained. These location coordinates are provided by the defect location information and record the start and end positions of the defective area on the fabric with millimeter-level accuracy.

[0095] The specific methods to improve scanning accuracy are as follows: First, reduce the fabric movement speed. This is achieved by sending a deceleration command to the production line control system, reducing the fabric movement speed from the current... meters per minute decreased to Meters per minute. With the speed reduced, the number of lines captured per unit length by the line scan camera in the light transmission imaging module increases to three times the original number, resulting in a corresponding increase in spatial sampling resolution. Second, increase the sampling frequency of the light transmission imaging module. Increase the camera's line scanning frequency from the current... Hertz increased to The Hertz rate was matched with the reduced fabric speed to ensure that the number of rows sampled per millimeter remained the same while the actual physical resolution was improved. Third, in yarn density measurement, a finer cylindrical lens array was used. Switching to cylindrical lenses with smaller lens unit widths (from 0.5 mm to 0.2 mm) allowed each yarn to correspond to more photoelectric pulse sampling points, thus improving the accuracy of density measurement. Fourth, in surface image acquisition, a higher resolution camera sensor or digital zoom was used to magnify the target area. The camera sensor pixel size was switched from 3.45 micrometers to 1.67 micrometers, or the lens focal length was switched from 25 mm to 50 mm, allowing the target area to occupy more pixels in the image. Fifth, in 3D reconstruction, the number of multi-view images acquired was increased. Additional mirrors in the mirror array were activated, increasing the number of viewing angles from 3 to 5, and multi-view fusion reduced the registration error of a single viewing angle.

[0096] An online fabric quality inspection device for textile production includes: The original design acquisition module is configured to acquire the original design of the digital printing pattern, wherein the original design includes information on each color channel and pixel resolution information. The light transmission imaging module is configured to emit detection light onto the fabric and collect the transmitted light signal that passes through the fabric, generate a multispectral light transmission image sequence of the fabric based on the transmitted light signal, and extract the light transmission feature values ​​of each spectral channel. The yarn density detection module is configured to measure the yarn density in the warp and weft directions respectively based on the alternating bright and dark spots corresponding to the warp and weft yarns in the transmitted light image, and calculate the fabric stretch rate by using the difference between the preset pixel period in the original design and the actual pixel period after printing. After dynamic tension calibration of the measurement results, the final warp yarn density and the final weft yarn density are output. The printing capacity calculation module is configured to calculate the basic printing capacity index based on the final warp yarn density, the final weft yarn density and the pixel resolution of the design original, and calculate the color correction coefficient based on the color channel information in the design original and the light transmission characteristic value of each spectral channel in the multispectral light transmission image sequence. The basic printing capacity index is dynamically corrected to obtain the final printing capacity index, and the light transmission uniformity of the light transmission image is calculated at the same time. The 3D detection module is configured to acquire surface images of the fabric from at least two different perspectives, reconstruct the three-dimensional morphology of the fabric surface based on the surface images, and extract a set of fabric defect parameters from the three-dimensional morphology. The quality assessment module is configured to calculate the fabric quality score based on the final printing capability index, light transmission uniformity, and fabric surface defect parameter set. When the fabric quality score is lower than a preset threshold, an unqualified signal is output.

[0097] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Any modifications, equivalent substitutions, and improvements made by those skilled in the art within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0098] In the description of this application, it should be noted that the terms "first", "second", and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0099] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0100] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the shown or discussed mutual couplings, direct couplings, or communication connections may be through some communication interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

[0101] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0102] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0103] Finally, it should be noted that the above-described embodiments are merely specific implementations of this application, used to illustrate the technical solutions of this application, and not to limit them. The protection scope of this application is not limited thereto. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some of the technical features, within the technical scope disclosed in this application. Such modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be covered within the protection scope of this application. Therefore, the protection scope of this application should be determined by the protection scope of the claims.

[0104] Furthermore, although the operations of the method of this application are described in a specific order in the accompanying drawings, this does not require or imply that these operations must be performed in that specific order, or that all the operations shown must be performed to achieve the desired result. Additionally or alternatively, certain steps may be omitted, multiple steps may be combined into one step, and / or one step may be broken down into multiple steps.

Claims

1. A method for online quality inspection of fabrics used in textile production, characterized in that, Includes the following steps: Obtain the original design artwork of the digital printing pattern, which includes information on each color channel and pixel resolution; Detection light is emitted onto the fabric and the transmitted light signal passing through the fabric is collected. A light transmission image of the fabric is generated based on the transmitted light signal. Based on the alternating bright and dark light spots corresponding to the warp and weft yarns in the transmitted light image, the number of yarns per unit length in the warp direction and the number of yarns per unit length in the weft direction of the fabric are measured respectively to determine the warp yarn density and the weft yarn density. The basic printing capacity index is calculated based on the warp yarn density, weft yarn density, and pixel resolution of the original design. The basic printing capacity index is then corrected based on the color channel information in the original design to obtain the final printing capacity index. At the same time, the light transmission uniformity of the transmitted image is calculated. The surface images of the fabric are acquired from at least two different perspectives. The three-dimensional morphology of the fabric surface is reconstructed from the surface images, and the set of fabric defect parameters is extracted from the three-dimensional morphology. The fabric quality score is calculated by comprehensively considering the final printing capacity index, light transmission uniformity, and fabric surface defect parameters. When the fabric quality score is lower than a preset threshold, an unqualified signal is output.

2. The online fabric quality detection method for textile production according to claim 1, characterized in that, When generating a translucent image, the following steps are included: The detection light is emitted onto the fabric surface in the form of parallel light, and the emission wavelength of the detection light is dynamically selected according to the fabric color. Transmitted light signals are collected on the side of the fabric facing away from the light source to generate a multispectral transmittance image sequence of the fabric. Transmission feature values ​​of each spectral channel in a multispectral transmission image sequence are extracted, and these transmission feature values ​​are used to calculate color correction coefficients.

3. The online fabric quality detection method for textile production according to claim 1, characterized in that, When measuring warp and weft yarn density, the following is included: The warp and weft interweaving structure of the fabric is transformed into a one-dimensional image of light and dark stripes through a cylindrical lens, so that the warp or weft yarns are clearly imaged rather than the yarn being measured is blurred. The light spot corresponding to each yarn in a one-dimensional bright and dark stripe image is converted into a photoelectric pulse signal; The original warp yarn density and the original weft yarn density were obtained by counting and measuring the period of the photoelectric pulse signal. The fabric stretch rate is calculated by using the difference between the preset pixel period in the original design and the actual pixel period after printing. Based on the fabric stretch rate, the original warp yarn density and the original weft yarn density are dynamically tension calibrated, and the final warp yarn density and the final weft yarn density are output.

4. The online fabric quality detection method for textile production according to claim 3, characterized in that, When calculating the basic printing capacity index, the following are included: Calculate the ratio of the final warp yarn density to the horizontal resolution of the original design, and the ratio of the final weft yarn density to the vertical resolution of the original design, respectively. The smaller of the two ratios is taken as the basic printing capacity index.

5. The online fabric quality detection method for textile production according to claim 2, characterized in that, The final printing capacity index is obtained by revising the basic printing capacity index, including: The color correction coefficient is calculated based on the proportion of each color channel in the original design and the light transmission characteristic values ​​of each spectral channel in the multispectral light transmission image sequence. The basic printing capability index described in claim 4 is dynamically corrected according to the color correction coefficient to obtain the final printing capability index. When the final printing capacity index is lower than the first threshold of the corresponding color, it is determined that the fabric yarn density is insufficient to support the pixel resolution of the original design, and an insufficient resolution warning signal is output.

6. The online fabric quality detection method for textile production according to claim 1, characterized in that, When calculating light transmittance uniformity, the following should be included: Calculate the ratio of the standard deviation to the mean of all pixel gray values ​​in the transmitted image; Calculate the difference between the number 1 and the ratio, and use it as the light transmission uniformity; When the light transmission uniformity is lower than the second threshold, an early warning signal for uneven yarn density distribution is output.

7. The online fabric quality detection method for textile production according to claim 1, characterized in that, When acquiring surface images of a fabric, the following are included: At least two cameras are used to simultaneously capture images of the fabric from the front and side views, respectively. After registering the multi-view images, three-dimensional point cloud data of the fabric surface is constructed. The following parameters are extracted from the 3D point cloud data: fluff density, fluff length, number of spherical defects, area of ​​spherical defects, height of spherical defects, wrinkle depth, wrinkle density, hole area, surface roughness, and surface smoothness.

8. The online fabric quality detection method for textile production according to claim 1, characterized in that, When inspecting the quality of printed fabric, the comprehensive fabric quality score includes: Obtain the geometric matching degree of the print; Weighting coefficients were assigned to the final printing capacity index, light transmittance uniformity, fabric defect parameter set, and printing geometric matching degree, respectively. The weighted sum of each item after assigning weight coefficients is used as the fabric quality score.

9. The online fabric quality detection method for textile production according to claim 1, characterized in that, After outputting an invalid signal, the following are included: Monitor the confidence level of the fabric quality score; When the confidence level is lower than the third threshold, the yarn density measurement, light transmission uniformity calculation and surface image acquisition are re-executed on the area corresponding to the fabric quality score with a higher scanning accuracy than the current resolution.

10. An online fabric quality inspection device for textile production, characterized in that, For performing the method according to any one of claims 1 to 9, comprising: The original design acquisition module is configured to acquire the original design of the digital printing pattern, wherein the original design includes information on each color channel and pixel resolution information. The light transmission imaging module is configured to emit detection light onto the fabric and collect the transmitted light signal that passes through the fabric, generate a multispectral light transmission image sequence of the fabric based on the transmitted light signal, and extract the light transmission feature values ​​of each spectral channel. The yarn density detection module is configured to measure the yarn density in the warp and weft directions respectively based on the alternating bright and dark spots corresponding to the warp and weft yarns in the transmitted light image, and calculate the fabric stretch rate by using the difference between the preset pixel period in the original design and the actual pixel period after printing. After dynamic tension calibration of the measurement results, the final warp yarn density and the final weft yarn density are output. The printing capacity calculation module is configured to calculate the basic printing capacity index based on the final warp yarn density, the final weft yarn density and the pixel resolution of the design original, and calculate the color correction coefficient based on the color channel information in the design original and the light transmission characteristic value of each spectral channel in the multispectral light transmission image sequence. The basic printing capacity index is dynamically corrected to obtain the final printing capacity index, and the light transmission uniformity of the light transmission image is calculated at the same time. The 3D detection module is configured to acquire surface images of the fabric from at least two different perspectives, reconstruct the three-dimensional morphology of the fabric surface based on the surface images, and extract a set of fabric defect parameters from the three-dimensional morphology. The quality assessment module is configured to calculate the fabric quality score based on the final printing capability index, light transmission uniformity, and fabric surface defect parameter set. When the fabric quality score is lower than a preset threshold, an unqualified signal is output.

Images