Image recognition-based cloth surface detection and spraying integrated device

Abstract

The application provides a cloth surface detection and spraying integrated equipment based on image recognition, and belongs to the technical field of cloth detection and spraying integrated equipment. The equipment comprises cloth storage rollers, a cloth spraying mechanism and cloth collection rollers which are sequentially arranged. The cloth surface detection and spraying integrated equipment based on image recognition further comprises a control unit, an image processing unit, a light source unit and an industrial camera. The image processing unit is used for performing pixel-level difference operation on a first image I_L and a second image I_C to obtain a difference image. Then, a pixel region with a gray value greater than an adaptive threshold in the difference image is determined as a geometric pseudo-defect region, and is shielded in subsequent color defect detection.

Description

Technical Field

[0001] This application relates to the technical field of fabric inspection and spraying equipment, and in particular to an integrated fabric surface inspection and spraying equipment based on image recognition. Background Technology

[0002] In textile printing, dyeing, and finishing production lines, fabrics are often continuously operated at high speeds of 60–120 m / min. Due to factors such as dye evenness, tension fluctuations, guide roller eccentricity, environmental lint, and uneven application of fluorescent whitening agents, the fabric surface is highly susceptible to defects such as color difference, fly waste, wrinkles, weft arcs, periodic roller marks, and localized deficiencies in fluorescent whitening agents. If these defects reach the finished product stage, they not only lead to downgrading and rework but also result in substantial claims and brand risks. Therefore, real-time detection and targeted repair must be carried out online.

[0003] Traditional manual inspection methods are labor-intensive, their detection rate is affected by subjective experience, and the human eye cannot operate stably in high-speed production line environments. Existing automated solutions mainly use two-dimensional CCD / CMOS cameras in conjunction with LED area array light sources for image acquisition, and then identify defects through thresholding or simple machine learning models. However, the imaging mode of a single angle and a single spectrum suffers from serious geometric pseudo-defect interference: wrinkles, flattened fibers, and fly waste adhere to the surface and appear as high-contrast shadows under low-angle light, which are often misjudged as true color differences, leading to mis-spraying by the sprayer.

[0004] Therefore, the industry urgently needs an integrated device that can operate at high speed and continuously, simultaneously detect geometric and chemical defects, has the ability to self-mask false defects, and can achieve millimeter-level closed-loop spraying repair, in order to replace manual labor and solve the above-mentioned technical pain points. Summary of the Invention

[0005] The purpose of this invention is to solve the above-mentioned problems by providing an integrated fabric surface inspection and spraying device based on image recognition.

[0006] The technical solution of this application is implemented as follows: In a first aspect, this application provides an integrated fabric surface inspection and spraying device based on image recognition. The integrated fabric surface inspection and spraying device based on image recognition includes a fabric storage roller, a fabric spraying mechanism, and a fabric collection roller. The fabric spraying mechanism is located between the fabric storage roller and the fabric collection roller. The effective spraying area of ​​the fabric spraying mechanism is the area to be sprayed. The fabric spraying mechanism includes a clamping component, a mobile control console, and a sprayer. The clamping component is used to clamp and fix the fabric in the area to be sprayed during spraying. The integrated fabric surface inspection and spraying equipment based on image recognition also includes: The system comprises a control unit, an image processing unit, a light source unit, and an industrial camera. The industrial camera is mounted above the area to be sprayed and faces the area to be sprayed. The image processing unit is signal-connected to a mobile control console and transmits the coordinates of the area to be sprayed to the mobile control console. The mobile control console moves the sprayer to the position of the area to be sprayed based on the coordinates of the area to be sprayed. The light source unit includes a low-angle light source and a coaxial light source. The control unit controls the industrial camera to acquire a first image I_L of the same area of ​​the fabric under the illumination of the low-angle light source and a second image I_C under the illumination of the coaxial light source. The image processing unit performs pixel-level difference operations on the first image I_L and the second image I_C to obtain a difference image. Then, pixel areas in the difference image with gray values ​​greater than an adaptive threshold are identified as geometric pseudo-defect areas and masked in subsequent color defect detection.

[0007] As a further improvement, the low-angle light source is an LED linear array, and the incident light forms a 25-35° tilt angle with the fabric surface.

[0008] As a further improvement, the coaxial light source is a ring-shaped LED and is coaxial with the industrial camera through a semi-transparent beam splitter to ensure shadowless illumination.

[0009] As a further improvement, the control unit includes an FPGA timing board, which generates two sets of strobe trigger signals TRIG_L and TRIG_C with a pulse width of 0.5 ms and an interval of 1.0 ms, ensuring that the fabric displacement of the two images is <0.05 mm.

[0010] As a further improvement, the image processing unit is used to perform pixel-level difference operations on the first image I_L and the second image I_C to obtain a difference image; then, the pixel regions in the difference image whose grayscale values ​​are greater than an adaptive threshold are determined as geometric pseudo-defect regions, specifically including: The image processing unit is used to perform 5x5 bilateral filtering (σ_color=15, σ_space=3) on I_L and I_C respectively to remove electronic noise while preserving wrinkle edges; Then, pixel-level differencing is performed as follows: ΔI(u,v)=|I_L(u,v) I_C(u,v)|, u∈[0,W-1], v∈[0,H-1], where W is the pixel width of the image and H is the pixel height of the image; The adaptive threshold is calculated as follows: Calculate the mean μ_Δ and standard deviation σ_Δ of the entire ΔI; base threshold T_base = μ_Δ + 2σ_Δ; introduce an illumination compensation factor k to compensate for the difference in reflectivity of the fabric itself: k = (mean(I_C) + 1) / (mean_ref + 1), where mean_ref = 128, 18% gray reference; obtain the final adaptive threshold, T_adapt = k·T_base, where k = 0.9~1.15; Binarization is determined as follows: M_fake(u,v)=1, if ΔI(u,v)>T_adapt, the pixel at coordinates (u,v) is determined to be a geometric pseudo-defect; M_fake(u,v)=0, otherwise, the pixel at coordinates (u,v) is not a geometric pseudo-defect.

[0011] To further improve the process, after the pixel at coordinates (u,v) is determined to be a geometric pseudo-defect, the image processing unit is further configured to: Remove isolated noise points using a 3x3 core. Retain areas with a size ∈ [50px, 8000px] to remove areas that are too large or too small; The preserved region is then subjected to Sobel again in the second image I_C. If the average gradient magnitude is less than 35 gray / ms, it is downgraded to "suspicious" instead of "confirmed" to reduce overkill.

[0012] As a further improvement, the light source unit further includes a 365nm narrowband ultraviolet LED; the control unit is used to control the industrial camera to acquire a third image I_uv of the same area of ​​the fabric under the illumination of the 365nm narrowband ultraviolet LED.

[0013] As a further improvement, the image processing unit is further configured to: Divide the third image I_uv and the second image I_C pixel by pixel: R = I_uv / (I_C+1); If R is more than 15% lower than the average of the same batch, it is judged as a color defect of "missing fluorescent whitening agent" and the coordinates are output.

[0014] As a further improvement, the image processing unit is further configured to: Perform a block-based fast Fourier transform on the second image I_C to obtain the power spectrum; The power spectrum is accumulated by angle in polar coordinates to obtain a power-angle histogram; If the half-width at half-maximum of the main peak in the histogram is greater than 8° and the main peak deviates from 90° by more than 2°, then a latitudinal arc defect is determined to exist, and the coordinates of the defect area extending along the latitudinal direction are output.

[0015] As a further improvement, the clamping assembly is mounted on the inner wall of the housing of the device corresponding to the area to be sprayed; The mobile control console is installed below the area to be painted. The sprayer is mounted on the mobile end of the mobile console. The sprayer moves in the area to be sprayed via the mobile console and sprays the fabric. An image marker is also mounted on the back of the sprayer. The image processing unit continuously analyzes the global image captured by the industrial camera, identifies and tracks the image marker, and thus obtains the current position of the sprayer on the fabric in real time. The integrated fabric surface inspection and coating equipment also includes: a redundant fabric roller group, which is movable up and down between the fabric storage roller and the fabric coating mechanism, and the fabric passes through the redundant fabric roller group.

[0016] The advantages or beneficial effects of the above technical solutions include at least the following: By integrating a deep learning-based image processing unit with a global vision industrial camera, the equipment can automatically, quickly, and accurately identify color defects on the fabric surface and generate precise coordinate information, completely replacing the traditional method that relies on human eye recognition. This ensures the objectivity and consistency of the inspection and adopts a "visual servo" control strategy. Attached Figure Description

[0017] The accompanying drawings illustrate exemplary embodiments of the present application and, together with the description thereof, serve to explain the principles of the present application. These drawings are included to provide a further understanding of the present application and are incorporated in and constitute a part of this specification.

[0018] Figure 1 A schematic diagram of the device according to an embodiment of the present invention is shown; Figure 2 The diagram shows a schematic of the device after one side panel is hidden according to an embodiment of the present invention, wherein part 2a is a schematic of the right side panel being hidden and part 2b is a schematic of the left side panel being hidden; Figure 3 A schematic diagram of a cylinder according to an embodiment of the present invention is shown; Figure 4 A schematic diagram of the mobile console structure according to an embodiment of the present invention is shown; Figure 5 The diagram illustrates a mobile control console moving a sprayer along the width of the fabric according to an embodiment of the present invention. Part 5a shows a diagram of the sprayer moving to one side of the width of the fabric, and part 5b shows a diagram of the sprayer moving to the other side of the width of the fabric. Figure 6The diagram illustrates how a mobile control console moves a sprayer along the length of a fabric according to an embodiment of the present invention. Part 6a shows a diagram of the sprayer moving to one side of the fabric length, and parts 6b show diagrams of the sprayer moving to the other side of the fabric length. Figure 7 A schematic diagram of a mobile console according to an embodiment of the present invention is shown; Figure 8 A schematic diagram of the spray nozzle according to an embodiment of the present invention is shown; Figure 9 A schematic diagram of the spring push rod according to an embodiment of the present invention is shown; Figure 10 A flowchart illustrating the workflow of an image recognition-based integrated fabric surface inspection and spraying device according to an embodiment of the present invention is shown.

[0019] Reference numerals: 10, Fabric storage roller; 20, Fabric spraying mechanism; 21, Cylinder; 22, Mobile control console; 221, Base; 2211, Bottom support frame; 2212, First motor; 2213, First rotating wheel; 2214, First transmission belt; 222, First guide rail; 223, First moving platform; 2231, Second motor; 2232, Second rotating wheel; 2233, Second transmission belt; 224, First... 225. Second guide rail; 226. Second moving stage; 227. Extension frame; 2261. Clearance groove; 2262. First fork arm; 2263. Second fork arm; 23. Sprayer; 231. Spraying hole; 30. Fabric redundant roller group; 31. First roller shaft; 32. Second roller shaft; 33. Sliding plate; 40. Fabric collecting roller; 50. Electric push rod; 60. Spring push rod; 61. Base; 62. Spring; 63. Telescopic rod. Detailed Implementation

[0020] Embodiments of this application will now be described in more detail with reference to the accompanying drawings. While some embodiments of this application are shown in the drawings, it should be understood that this application can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this application. It should be understood that the drawings and embodiments of this application are for illustrative purposes only and are not intended to limit the scope of protection of this application.

[0021] It should be noted that, where there is no conflict, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0022] It should be understood that the term "comprising" and its variations as used herein are open-ended, meaning "including but not limited to". The term "based on" means "at least partially based on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". Definitions of other terms will be given in the following description. It should be noted that the concepts of "first", "second", etc., mentioned in this application are used only to distinguish different devices, modules, or units, and are not intended to limit the order of functions performed by these devices, modules, or units or their interdependencies.

[0023] It should be noted that the terms "one" and "more" used in this application are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0024] The names of the messages or information exchanged between multiple devices in the embodiments of this application are for illustrative purposes only and are not intended to limit the scope of these messages or information.

[0025] Reference Figure 1 This invention provides an integrated fabric surface inspection and spraying device based on image recognition. The device includes a fabric storage roller 10 for storing fabric, a fabric spraying mechanism 20, and a fabric collection roller 40. The fabric sequentially travels from the fabric storage roller 10 to the fabric spraying mechanism 20 for spraying and then enters the fabric collection roller 40. The rotating fabric storage roller 10 continuously outputs fabric to the fabric spraying mechanism 20. The effective spraying area of ​​the fabric spraying mechanism 20 is the area to be sprayed.

[0026] The fabric spraying mechanism 20 includes: The clamping assembly is installed on the inner wall of the area to be sprayed near the fabric. During spraying, the clamping assembly clamps and fixes the fabric in the area to be sprayed. Specifically, the clamping assembly includes a cylinder 21. The base 61 end of the cylinder 21 is installed on the inner wall of the equipment, and a rubber pad is installed on the telescopic end. The mobile control console 22 is installed below the area to be sprayed. The mobile control console 22 includes a base 221 installed below the area to be sprayed. The clamping assembly further includes a bottom support 2211, which is installed on the base 221. The position of the bottom support 2211 corresponds to the rubber pad. When the cloth enters the area to be sprayed, the telescopic end of the cylinder 21 extends and fixes the cloth to the bottom support 2211 through the rubber pad. The sprayer 23 is mounted on the mobile control console 22. The sprayer 23 moves in the area to be sprayed via the mobile control console 22 and detects the color of the fabric at different positions in the area to be sprayed. The image recognition-based integrated fabric surface inspection and spraying device further includes a control unit, an image processing unit, a light source unit, and an industrial camera (not shown in the figure). The industrial camera is installed above the area to be sprayed and faces the area to be sprayed to take pictures. The image processing unit is signal-connected to the mobile console and transmits the coordinates of the area to be sprayed to the mobile console. The mobile console moves the sprayer to the position of the coordinates of the area to be sprayed based on the coordinates of the area to be sprayed. The light source unit includes a low-angle light source and a coaxial light source. The control unit is used to control the industrial camera to acquire a first image I_L of the same area of ​​the fabric under the illumination of the low-angle light source and a second image I_C under the illumination of the coaxial light source.

[0027] As a further improvement, the low-angle light source is an LED linear array, and the incident light rays form a 25-35° angle with the fabric surface, and the height from the fabric surface is 50-150mm. The setting of the low-angle light source is existing technology and will not be described further here.

[0028] As a further improvement, the coaxial light source is a ring-shaped LED, and it is coaxial with the industrial camera through a semi-transparent beam splitter to ensure shadowless illumination. The configuration of the coaxial light source is existing technology and will not be described further here.

[0029] As a further improvement, the control unit includes an FPGA timing board, which generates two sets of strobe trigger signals TRIG_L and TRIG_C with a pulse width of 0.5 ms and an interval of 1.0 ms, ensuring that the fabric displacement of the two images is <0.05 mm.

[0030] Please see also Figure 10 The collaborative workflow is as follows: First, the industrial camera sends the captured image of the fabric to the image processing unit. The image processing unit analyzes the fabric image, identifies the location of color defects in the fabric, and generates the corresponding coordinates of the area to be sprayed in the coordinate system of the fabric to be sprayed. Subsequently, the image processing unit sends this coordinate information to the mobile control console 22. Based on the received coordinates, the mobile control console 22 drives the sprayer 23 to move towards the target position and spray.

[0031] The image processing unit analyzes the image and identifies the location of color defects in the fabric, specifically including: The image processing unit is used to perform pixel-level difference operations on the first image I_L and the second image I_C to obtain a difference image; then, the pixel regions in the difference image with gray values ​​greater than an adaptive threshold are identified as geometric pseudo-defect regions and are masked in subsequent color defect detection.

[0032] In one embodiment, the image processing unit is configured to perform pixel-level difference operations on the first image I_L and the second image I_C to obtain a difference image; then, the pixel regions in the difference image whose grayscale values ​​are greater than an adaptive threshold are determined as geometric pseudo-defect regions, specifically including: The image processing unit is used to perform 5x5 bilateral filtering on I_L and I_C respectively (in one embodiment, the color / grayscale standard deviation σ_color is 15 and the spatial standard deviation σ_space is 3) to remove electronic noise while preserving wrinkle edges; Then, pixel-level differencing is performed as follows: ΔI(u,v)=|I_L(u,v) I_C(u,v)|, u∈[0,W-1], v∈[0,H-1], where W is the pixel width of the image and H is the pixel height of the image; The adaptive threshold is calculated as follows: Calculate the mean μ_Δ and standard deviation σ_Δ of the entire ΔI; base threshold T_base = μ_Δ + 2σ_Δ; introduce an illumination compensation factor k to compensate for the difference in reflectivity of the fabric itself: k = (mean(I_C) + 1) / (mean_ref + 1), where mean_ref = 128 (18% gray baseline); obtain the final adaptive threshold, T_adapt = k·T_base, where k = 0.9~1.15, k≈0.9~1.15, T_adapt falls in the 18~28 gray (grayscale) range, thus covering more than 95% of wrinkle / fly pattern samples; Binarization is determined as follows: M_fake(u,v)=1, if, ΔI(u,v)>T_adapt; M_fake(u,v)=0, otherwise.

[0033] M_fake is a binary mask of the same size as the original image, where each pixel takes only 0 or 1. M_fake(u,v)=1 indicates that the pixel at coordinates (u,v) is judged as a geometric pseudo-defect (wrinkle, flyaway, shadow, etc.), and this area will be masked in subsequent color defect detection; If M_fake(u,v)=0, it means that the pixel is not a geometric pseudo-defect and can proceed to the next step of real color defect detection.

[0034] In other embodiments, as a further improvement, after the pixel at coordinates (u,v) is determined to be a geometric pseudo-defect, the image processing unit is further configured to: Remove isolated noise points using a 3x3 core. Retain areas with a size ∈ [50px, 8000px] to remove areas that are too large or too small; The preserved region is then subjected to Sobel again in the second image I_C. If the average gradient magnitude is less than 35 gray / ms, it is downgraded to "suspicious" instead of "confirmed" to reduce overkill.

[0035] Sobel is a first-order differential edge detection operator used to quickly calculate the gray-level gradient (i.e., the rate of change) of each pixel in an image. The larger the gradient, the more pronounced the edge.

[0036] To achieve precise positioning of the sprayer 23, an image marker is installed on the back of the sprayer 23. The image processing unit continuously analyzes the global image captured by the industrial camera, identifies and tracks the image marker, thereby obtaining the current position of the sprayer 23 on the fabric in real time. When the image processing unit determines that the image marker in the image coincides with the coordinate position of the area to be sprayed, it confirms that the sprayer 23 has moved to the target position and the spraying operation can be started.

[0037] The image processing unit integrates a trained deep learning model for fabric defect detection.

[0038] In one embodiment, as a further improvement, the light source unit further includes a 365nm narrowband ultraviolet LED; the control unit is used to control the industrial camera to acquire a third image I_uv of the same area of ​​the fabric under the illumination of the 365nm narrowband ultraviolet LED.

[0039] Localized deficiencies in fluorescent whitening agent (FWA) can cause fabric to appear "dark yellow" under store lighting, but conventional RGB cameras cannot detect this in advance. To address this issue, in one embodiment, as a further improvement, the image processing unit is further configured to: Divide the third image I_uv and the second image I_C pixel by pixel, resulting in R = I_uv / (I_C+1). This pixel-by-pixel division of the ultraviolet image (I_uv) and the second image I_C yields a new "ratio image" R, used to highlight areas containing fluorescent whitening agents. The third image I_uv is taken under a 365 nm ultraviolet lamp; areas with more fluorescent whitening agents have higher pixel values. The second image I_C(u,v) is taken under ordinary visible light—reflecting the overall brightness of the fabric. Adding 1 (I_C+1) to the denominator prevents division by zero (in black areas). If R is more than 15% lower than the average of the same batch, it is judged as a color defect of "lack of fluorescent whitening agent" and the coordinates are output. The larger the result R is, the more "strong ultraviolet light and weak visible light" it is, that is, the high content of fluorescent whitening agent; if R is significantly smaller, it may be a deficiency of whitening agent (defect).

[0040] When the skew of the weft yarn in warp-knitted fabric is greater than 2°, "shadow stripes" are easily produced during subsequent setting, which cannot be identified by traditional methods. To solve the above problem, in one embodiment, as a further improvement, the image processing unit is further used for: Perform a block-based fast Fourier transform on the second image I_C to obtain the power spectrum; The power spectrum is accumulated by angle in polar coordinates to obtain a power-angle histogram; If the half-width at half-maximum of the main peak of the histogram is greater than 8° and the main peak deviates from 90° by more than 2°, then a latitudinal arc defect is determined to exist, and the coordinates of the defect area extending along the latitudinal direction are output, thereby solving the implicit color difference of the "latitudinal slant / arc slant" type.

[0041] Roller eccentricity or gearbox meshing errors can leave faint color variations on the fabric surface with a spacing of 1-2 mm, which are difficult to detect with the naked eye but are easily downgraded during finished product inspection. To solve the above problem, in one embodiment, as a further improvement, the image processing unit is further used for: The grayscale curve is extracted along the length of the fabric, and the amplitude spectrum is obtained by performing a one-dimensional Fourier transform on the grayscale curve. If the amplitude spectrum is between 5 and 50 mm -1 If there are three or more equally spaced peaks within the spatial frequency range and the peak-to-valley difference is greater than 3gray, then a periodic color defect is determined to exist, and the corresponding phase segment coordinates are output.

[0042] To convert the bounding box position into coordinates that the sprayer can execute, the system needs to be calibrated during installation and commissioning: a calibration plate with known physical dimensions, such as a checkerboard pattern, is placed stationary in the area to be sprayed, and the image pixel coordinates are captured and identified by an industrial camera to establish a precise mapping relationship between the image pixel coordinates and the mechanical coordinates of the two-dimensional guide rail of the mobile console.

[0043] During actual operation, the image processing unit converts the pixel coordinates of the center point of the detected defect bounding box into the target two-dimensional coordinates (X1, Y1) in the coordinate system of the mobile console in real time through the above mapping relationship. Together, they form the two-dimensional coordinates (X2, Y2) of the area to be sprayed on the fabric, and are sent to the mobile console to drive the sprayer to position.

[0044] The specific steps are as follows: After the fabric is clamped and secured, the equipment automatically executes the following cycle to process all identified defects sequentially: S1. Defect identification and target coordinate generation: Action: The industrial camera captures high-resolution images of the static fabric surface and transmits them to the image processing unit.

[0045] Processing: The deep learning model built into the image processing unit analyzes the image, identifies color defect areas, and outputs their bounding box information; Conversion: The system selects the center point of the defect area and uses the pre-calibrated mapping relationship to convert it into a precise two-dimensional target coordinate (X_target, Y_target) in the coordinate system of the mobile console, where X corresponds to the width direction of the fabric and Y corresponds to the length direction of the fabric.

[0046] S2. Obtain the current position of the sprayer: Action: The system continuously captures a global view using an industrial camera, identifying and tracking image markers mounted on the back of the sprayer; Processing: The image processing unit calculates the position of the image marker in the current screen in real time, and obtains the current mechanical coordinates (X_current, Y_current) of the sprayer through the same coordinate mapping relationship.

[0047] S3. Coordinate difference calculation and motion decision: Calculation: The controller calculates the difference between the target coordinates and the current coordinates. ΔX = X_target - X_current; ΔY = Y_target - Y_current; Decision: Compare ΔX and ΔY with the system's set positioning tolerance threshold (e.g., ±1 mm): If |ΔX|>threshold, then it is determined that the first motor needs to be started to drive the sprayer to move along the X-axis.

[0048] If |ΔY|>threshold, then it is determined that the second motor needs to be started to drive the sprayer to move along the Y-axis.

[0049] If the difference between the two values ​​is less than the threshold, it is determined that the sprayer has reached the target position.

[0050] S4, Dual-axis cooperative drive and real-time tracking: Drive: The controller sends pulse commands to the first motor and / or the second motor according to the values ​​and signs of ΔX and ΔY, driving the first moving stage (X direction) and the second moving stage (Y direction) to move; The two axes can move independently or simultaneously, allowing the sprayer to move toward the target point along the shortest path or a specific trajectory.

[0051] Closed-loop feedback: During the movement, S2 continues, and the system continuously updates the coordinates (X_current, Y_current) of the sprayer and recalculates the difference ΔX and ΔY to form real-time position feedback.

[0052] S5. Confirmation of Location and Execution of Spraying: Judgment: When both |ΔX| and |ΔY| calculated in real time are less than the tolerance threshold, the system determines that the positioning is complete.

[0053] Execution: The controller sends an start command to the sprayer to perform precise spraying. After spraying is completed, the system returns to step S1 to process the next defect coordinate, until all defects in the current screen have been processed.

[0054] The equipment also includes: a fabric redundancy roller group 30, which is movably installed between the fabric storage roller 10 and the fabric spraying mechanism 20. When the fabric passes through the fabric redundancy roller group 30, the fabric needs to temporarily stay in the spraying area when the sprayer 23 is spraying. Since the rotating fabric storage roller 10 continuously outputs fabric to the fabric spraying mechanism 20, the fabric will form redundancy between the fabric spraying mechanism 20 and the fabric storage roller 10. When the clamping component clamps the fabric in the area to be sprayed, the fabric redundancy roller group 30 moves upward, so that the fabric forms a redundant part. When the clamping component releases the clamping component from the fabric, the fabric redundancy roller group 30 moves downward, so that the redundant part of the fabric enters the area to be sprayed.

[0055] On both sides of the area to be sprayed, there are two rollers used to work with the clamping assembly to stretch the fabric.

[0056] Because the fabric needs to remain temporarily in the spraying area during the coating process, without the redundant fabric roller group 30, the fabric storage roller 10 would need to stop rotating before coating. The accumulation of redundancy allows the fabric storage roller 10 to rotate without stopping. Figure 2 As shown in parts 2a and 2b.

[0057] In the above, the redundant fabric roller group 30 includes: two sliding plates 33, which are respectively installed inside the equipment and can be moved up and down; a first roller shaft 31, which is rotatably mounted on the sliding plates 33 at both ends; and a second roller shaft 32, which is rotatably mounted on the sliding plates 33 at both ends and located below the first roller shaft 31. The portion of the fabric that passes through the fabric storage roller 10 passes between the first roller shaft 31 and the second roller shaft 32. When the sliding plate 33 moves upward, the first roller shaft 31 and the second roller shaft 32 will pull the fabric upward. When the sliding plate 33 moves downward, it will send the redundant portion of the fabric to the fabric spraying mechanism 20. Based on the further improvements to the above structure, the mobile console 22 includes: Base 221 is installed below the area to be painted; The first guide rail 222 is mounted on the upper surface of the base 221 and extends along the width of the fabric. The first moving stage 223 is movably mounted on the first guide rail 222 via the first drive assembly, and the axis of the first moving stage 223 is consistent with the direction of fabric travel. Among them, such as Figure 7 As shown, the first drive assembly includes: a first motor 2212, the body of which is mounted on a base 221; a first wheel 2213, which is rotatably mounted on the base 221 via a bracket; a first transmission belt 2214, which is sleeved on the output end of the first motor 2212 and the first wheel 2213; a first moving platform 223 is connected to the first transmission belt 2214, wherein the axial direction of the first transmission belt 2214 is consistent with the axial direction of the first guide rail 222; the rotation of the output end of the first motor 2212 causes the first transmission belt 2214 to move, thereby moving the first moving platform 223 on the base 221 along the width direction of the fabric. The mobile control console 22 further includes: a second guide rail 224, mounted on the upper surface of the first mobile stage 223 and extending along the axial direction of the first mobile stage 223; a second mobile stage 225, movably mounted on the second guide rail 224 via a second drive assembly, the second drive assembly including: a second motor 2231, the motor body of which is mounted on the base 221; a second wheel 2232, rotatably mounted on the first mobile stage 223 via a bracket; and a second transmission belt 2233, sleeved on the output end of the second motor 2231 and the second wheel 2232; the second mobile stage 225 is connected to the second transmission belt 2233, wherein the axial direction of the second transmission belt 2233 is aligned with the axial direction of the second guide rail 224; the rotation of the output end of the second motor 2231 causes the second transmission belt to move, thereby moving the second mobile stage 225 along the length of the fabric on the first mobile stage 223.

[0058] Furthermore, an extension frame 226 is installed on the second moving platform 225. The first end of the extension frame 226 in the length direction has a clearance groove 2261 extending towards the second end. The clearance groove 2261 causes the first end of the extension frame 226 to form a first fork arm 2262 and a second fork arm 2263. The fabric is located in the clearance groove 2261, with the first fork arm 2262 above the fabric and the second fork arm 2263 below the fabric. A sprayer 23 is installed at the end of the first fork arm 2262, and the spray nozzle 231 is arranged downwards. Figure 8 As shown; like Figure 5 As shown in sections 5a and 5b, the output of the first motor 2212 rotates, driving the first moving platform 223 to move back and forth, causing the sprayer 23 to move from one side (section 5a) to the other side (section 5b) along the width of the fabric. Figure 6As shown in parts 6a and 6b, the output end of the second motor 2231 rotates, driving the second moving platform 225 to move on the first moving platform 223, so that the sprayer 23 moves from one side (part 6a) to the other side (part 6b) along the length of the fabric. The movement of the first moving platform 223 and the second moving platform 225 do not interfere with each other, and the sprayer 23 can be adjusted to any position in the area to be sprayed for spraying work.

[0059] Based on the further improvement of the above structure, the device also includes a controller; the controller is installed in the device and connected to the cylinder 21, and the controller is used to control the extension and retraction of the telescopic end of the cylinder 21 periodically. The device also includes: a first drive motor for rotating the fabric storage roller 10, the first drive motor being installed on one side of the device located on the fabric storage roller 10 (not shown in the figure); a second drive motor for rotating the fabric collection roller 40 (not shown in the figure), the second drive motor being installed on one side of the device located on the fabric collection roller 40, the second drive motor being a servo motor; both the first drive motor and the second drive motor are connected to the controller; when the controller controls the extension end of the cylinder 21 to extend, it controls the second drive motor to stop rotating to prevent the fabric from being torn.

[0060] Furthermore, the sliding plate 33 has a connecting plate in the middle, and the connecting plate passes through the side wall of the equipment and extends out of the outside of the equipment. An electric push rod 50 is installed on the outside of the equipment. The base 61 end of the electric push rod 50 is installed on the outside of the equipment, and the telescopic end is fixed to the connecting plate. The controller is also connected to the electric push rod 50. The controller can control the extension and retraction of the extension end of the electric push rod 50. When the cylinder 21 clamps the cloth in the area to be sprayed, the controller controls the extension end of the electric push rod 50 to extend, driving the redundant cloth roller group 30 to move upward, so that the cloth forms a redundant part. When the cylinder 21 releases the clamping of the cloth, the controller controls the extension end of the electric push rod 50 to retract, driving the redundant cloth roller group 30 to move downward, so that the redundant cloth enters the area to be sprayed.

[0061] As an alternative to the above embodiment: the middle part of the sliding plate 33 has a connecting plate, and the connecting plate passes through the side wall of the equipment and extends out of the outside of the equipment; a spring push rod 60 is also installed on the outside of the equipment, the base 61 end of the spring push rod 60 is installed on the outside of the equipment (not shown in the figure), and the telescopic end is fixed to the connecting plate, wherein the base 61 end of the spring push rod 60 is located below the connecting plate, and the telescopic end can extend upward elastically; like Figure 9 As shown, the spring push rod 60 includes a base 61, a telescopic rod 63, and a spring 62 installed in the base 61. One end of the spring 62 abuts against the inner bottom surface of the base 61, and the other end abuts against the rear end of the telescopic rod 63.

[0062] In this embodiment, when the cylinder 21 clamps the fabric in the area to be sprayed, the fabric storage roller 10 rotates, conveying the fabric between the fabric spraying mechanism 20 and the fabric storage roller 10. The spring 62 inside the spring push rod 60 loses its downward pressure, and thus extends upward due to its elastic deformation, driving the sliding plate 33 to move upward through the connecting plate, forming a redundant part of the fabric. When the cylinder 21 releases the clamp on the fabric, the second drive motor rotates, pulling the fabric. When the fabric is pulled, it tends to be taut between the spraying area and the fabric storage roller 10. This taut state causes the fabric to be positioned on the second roller. A portion of shaft 32 forms a downward pressure, acting on the upper surface of the second roller shaft 32, thereby pulling the entire redundant fabric roller group 30 downward to send the redundant part of the fabric into the fabric spraying mechanism 20 while avoiding delays in fabric transportation. When cylinder 21 clamps the fabric in the area to be sprayed, the fabric storage roller 10 rotates, conveying the fabric between the fabric spraying mechanism 20 and the fabric storage roller 10. The spring 62 in the spring push rod 60 loses the downward pressure, and thus extends upward due to the elastic deformation of the backtracking. It drives the sliding plate 33 to move upward through the connecting plate, forming the redundant part of the fabric in this reciprocating cycle.

[0063] Compared to the electric push rod 50, this embodiment does not require precise control of the fabric redundant roller group 30, but utilizes the natural tension and relaxation state of the fabric to create redundancy.

[0064] In the description of this application, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this application.

[0065] Those skilled in the art should understand that the above embodiments are merely for illustrative purposes and are not intended to limit the scope of this application. Those skilled in the art can make other changes or modifications based on the above disclosure, and these changes or modifications still fall within the scope of this application.

Claims

1. An integrated fabric surface inspection and spraying device based on image recognition, characterized in that: The device includes a fabric storage roller, a fabric spraying mechanism, and a fabric collecting roller. The fabric spraying mechanism is located between the fabric storage roller and the fabric collecting roller. The effective spraying area of ​​the fabric spraying mechanism is the area to be sprayed. The fabric spraying mechanism includes a clamping component, a mobile control console, and a sprayer. The clamping component is used to clamp and fix the fabric in the area to be sprayed during spraying. The integrated fabric surface inspection and spraying equipment based on image recognition also includes: The system comprises a control unit, an image processing unit, a light source unit, and an industrial camera. The industrial camera is mounted above the area to be sprayed and faces the area to be sprayed. The image processing unit is signal-connected to a mobile control console and transmits the coordinates of the area to be sprayed to the mobile control console. The mobile control console moves the sprayer to the position of the area to be sprayed based on the coordinates of the area to be sprayed. The light source unit includes a low-angle light source and a coaxial light source. The control unit controls the industrial camera to acquire a first image I_L of the same area of ​​the fabric under the illumination of the low-angle light source and a second image I_C under the illumination of the coaxial light source. The image processing unit performs pixel-level difference operations on the first image I_L and the second image I_C to obtain a difference image. Then, pixel areas in the difference image with gray values ​​greater than an adaptive threshold are identified as geometric pseudo-defect areas and masked in subsequent color defect detection. The low-angle light source is an LED linear array, and the incident light forms a 25-35° tilt angle with the fabric surface; the coaxial light source is a ring LED, and is coaxial with the industrial camera through a semi-transparent beam splitter to ensure shadow-free illumination; the light source unit further includes a 365nm narrowband ultraviolet LED; the control unit is used to control the industrial camera to acquire a third image I_uv of the same area of ​​the fabric under the illumination of the 365nm narrowband ultraviolet LED; The image processing unit is used to perform pixel-level difference operations on the first image I_L and the second image I_C to obtain a difference image; then, the pixel regions in the difference image whose grayscale values ​​are greater than an adaptive threshold are determined as geometric pseudo-defect regions, specifically including: The image processing unit is used to perform 5x5 bilateral filtering on the first image I_L and the second image I_C respectively, wherein the color / grayscale standard deviation σ_color is 15 and the spatial standard deviation σ_space is 3, in order to remove electronic noise while retaining wrinkled edges; Then, pixel-level differencing is performed as follows: ΔI(u,v)=|I_L(u,v) I_C(u,v)|, u∈[0,W-1], v∈[0,H-1], where W is the pixel width of the image and H is the pixel height of the image; The adaptive threshold is calculated as follows: Calculate the mean μ_Δ and standard deviation σ_Δ of the entire ΔI; base threshold T_base = μ_Δ + 2σ_Δ; introduce an illumination compensation factor k to compensate for the difference in reflectivity of the fabric itself: k = (mean(I_C) + 1) / (mean_ref + 1), where mean_ref = 128, 18% gray reference; obtain the final adaptive threshold, T_adapt = k·T_base, where k = 0.9~1.15; Binarization is determined as follows: M_fake(u,v)=1, if ΔI(u,v)>T_adapt, the pixel at coordinates (u,v) is determined to be a geometric pseudo-defect; M_fake(u,v)=0, otherwise, the pixel at coordinates (u,v) is not a geometric pseudo-defect; The image processing unit is further configured to: Divide the third image I_uv and the second image I_C pixel by pixel: R = I_uv / (I_C+1); If R is more than 15% lower than the average of the same batch, it is judged as a color defect of "missing fluorescent whitening agent" and the coordinates are output.

2. The integrated fabric surface inspection and spraying equipment based on image recognition according to claim 1, characterized in that: The control unit includes an FPGA timing board, which generates two sets of strobe trigger signals TRIG_L and TRIG_C with a pulse width of 0.5ms and an interval of 1.0ms, ensuring that the fabric displacement of the two images is <0.05 mm.

3. The integrated fabric surface inspection and spraying equipment based on image recognition according to claim 1, characterized in that: After the pixel at coordinates (u,v) is determined to be a geometric pseudo-defect, the image processing unit is further configured to: Remove isolated noise points using a 3x3 core. Retain areas with a size ∈ [50px, 8000px] to remove areas that are too large or too small; The preserved region is subjected to Sobel again in the second image I_C. If the average gradient magnitude is less than 35 gray / ms, it is downgraded to "suspicious" instead of "confirmed" to reduce overkill.

4. The integrated fabric surface inspection and spraying equipment based on image recognition according to claim 1, characterized in that: The image processing unit is further configured to: Perform a block-based fast Fourier transform on the second image I_C to obtain the power spectrum; The power spectrum is accumulated by angle in polar coordinates to obtain a power-angle histogram; If the half-width at half-maximum of the main peak in the histogram is greater than 8° and the main peak deviates from 90° by more than 2°, then a latitudinal arc defect is determined to exist, and the coordinates of the defect area extending along the latitudinal direction are output.

5. The integrated fabric surface inspection and spraying equipment based on image recognition according to claim 4, characterized in that: The clamping assembly is installed on the inner sidewall of the outer casing of the device corresponding to the area to be sprayed. The mobile control console is installed below the area to be painted. The sprayer is mounted on the mobile end of the mobile console. The sprayer moves in the area to be sprayed via the mobile console and sprays the fabric. An image marker is also mounted on the back of the sprayer. The image processing unit continuously analyzes the global image captured by the industrial camera, identifies and tracks the image marker, and thus obtains the current position of the sprayer on the fabric in real time. The integrated fabric surface inspection and coating equipment also includes: a redundant fabric roller group, which is movable up and down between the fabric storage roller and the fabric coating mechanism, and the fabric passes through the redundant fabric roller group.

Images