Online detection method for defects of glass fiber gridding cloth based on machine vision

Abstract

The invention relates to the technical field of glass fiber gridding cloth defect detection, and discloses a glass fiber gridding cloth defect online detection method based on machine vision. Comprising the steps of multi-channel high-speed stroboscopic light source arrangement, camera deployment, motion platform conveyor belt construction, microscope lens construction, image preprocessing, defect detection and feature extraction, defect real-time feedback system construction and data management. By means of the detection method, the defect detection effect of the glass fiber gridding cloth can be improved through multi-dimensional high-precision detection.

Description

Technical Field

[0001] The present invention belongs to the technical field of processing for detecting defects in glass fiber mesh cloth, and in particular relates to an online method for detecting defects in glass fiber mesh cloth based on machine vision. Background Art

[0002] Fiberglass mesh is a composite material based on woven fiberglass fabric, coated with a polymer alkali-resistant emulsion. It features high tensile strength, excellent alkali resistance, and good flexibility. Its high tensile strength in both warp and weft directions makes it widely used in building interior and exterior wall insulation, waterproofing, fire protection, and crack prevention projects, enhancing structural stability and preventing wall cracking and shedding.

[0003] Poor-quality mesh cloth (such as clay crucible wire drawing products made from waste glass) is prone to pulverization and failure in alkaline environments, causing wall cracks and even structural safety hazards; mesh cloth with insufficient alkali resistance will experience a sharp drop in tensile strength in a cement mortar environment, losing its reinforcement effect.

[0004] Traditional manual inspection has limitations. Relying on visual inspection or laboratory sampling (such as tensile strength testing), it cannot achieve full-width, continuous inspection, resulting in a high rate of missed detections. Furthermore, manual operation is slow and cannot meet the real-time quality monitoring needs of production lines. Some existing inspection equipment, such as handheld thickness gauges, only tests a single parameter and lacks the ability to conduct multi-dimensional defect analysis. They are unable to detect micron-level defects (such as broken edges and white lines), resulting in hidden quality issues going undetected and poor inspection results.

[0005] To this end, we propose an online detection method for glass fiber mesh defects based on machine vision. Summary of the Invention

[0006] The purpose of the present invention is to provide an online detection method for glass fiber mesh defects based on machine vision in order to improve the detection effect of glass fiber mesh defects.

[0007] The technical solution adopted in the present invention is as follows: A method for online detection of defects in glass fiber mesh cloth based on machine vision, wherein the method is as follows: S1: A multi-channel, high-speed stroboscopic light source arrangement uses a red and blue LED light source mounted directly above the fabric or at an angle of 15°-75° to the inspection surface. Time-sharing stroboscopic light (≥1kHz) is used to achieve three-dimensional layered imaging of scratches and oil stains on the fiber surface. A white LED backlight is mounted below the fabric to detect translucent defects such as splits and cracks. The defective area becomes a highlight due to the light transmission. S2: Camera deployment: A high-resolution line scan camera is installed perpendicular to the fabric surface, directly above the motion platform, covering a width of 1-4 meters to ensure global scanning without blind spots. Auxiliary line scan cameras are installed at an angle of 30°-45° to the side of the fabric's travel direction to enhance the ability to capture directional defects such as broken warps and wefts. S3: The motion platform conveyor belt is constructed. A servo motor drives a precision roller conveyor belt. The fabric is laid flat on the conveyor belt surface. The speed stability is ±0.05%, and the vibration amplitude is less than 5μm, ensuring strict synchronization with the camera line frequency. The encoder is installed on the conveyor belt drive shaft to provide real-time feedback of motion parameters, triggering the camera to collect and compensate for speed fluctuations. S4: Microscope lens installation: A 5x microscope lens is installed on the side of the motion platform. Combined with an area scan camera, it performs secondary imaging of suspected defect areas, enabling local magnification detection of subtle defects such as micron-level broken edges and white lines. S5: Image preprocessing: noise removal and image segmentation are performed on the captured images. Color images are converted to grayscale images to reduce data dimensionality and unify feature representation. The captured images are smoothed using a mean filter to eliminate random noise caused by mechanical vibration. During image segmentation, a global threshold method is applied to achieve binary segmentation of the fiber substrate and defects. Defective areas are marked as white pixel blocks. High-frequency edge information is then enhanced to improve the continuity characteristics of linear defects such as scratches and cracks. S6: Defect detection and feature extraction. Close operations are used to eliminate interfering noise points with an area less than 0.2 mm², retaining valid defect contours. The findContours function is used to extract defect geometric parameters (such as area ≥ 0.1 mm² and length ≥ 4.25 mm) for classification. When detecting bright spot defects, the system uses specular reflection characteristics to locate them by using a sudden drop in grayscale value (Δ> 60). When detecting broken warp and weft, the Hough transform is used to detect the angular offset of the warp and weft breaks (≥ 15° is considered an abnormality). When detecting foreign matter contamination, the HSV color space is used to analyze the saturation abnormality of the color spot area (S value > 120). S7: Real-time defect feedback system is built, using RS485 communication protocol, triggering sound and light alarms within 50ms after a defect is detected, and simultaneously locating the defect coordinates; S8: Data management: Establish an SQL database to store the defect map, location coordinates and processing records of each roll of cloth to support data backtracking, and use the intelligent cloud platform to compile statistical defect distribution heat maps to analyze process parameters and defect rates.

[0008] In a preferred embodiment of the invention, the multi-channel high-speed stroboscopic light source adopts a red and blue dual-color LED light source.

[0009] In a preferred embodiment of the invention, the red light channel (wavelength 630nm) in the multi-channel high-speed stroboscopic light source is used to detect surface foreign matter and hair balls, and the blue light channel (wavelength 450nm) enhances the transmitted light to capture internal defects such as splits and cracks.

[0010] In a preferred embodiment of the invention, the real-time defect feedback system includes a PLC controller, an audible and visual alarm, and a repeater.

[0011] In a preferred embodiment of the invention, the high-resolution linear array camera collects mesh cloth images in real time, detects defects, and sends defect data to the PLC controller via RS485. The PLC controller parses the data, records defect information, and triggers an alarm instruction. After receiving the instruction, the sound and light alarm starts the sound and light prompt to achieve real-time feedback.

[0012] In a preferred embodiment of the invention, the repeater is used to dynamically adjust the signal strength to ensure the stability of long-distance communications.

[0013] In a preferred embodiment of the invention, the high-resolution line array camera is an 8k resolution line array camera (pixel size 5 μm) equipped with a telecentric lens (FOV 1.2 meters).

[0014] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are: 1. This invention utilizes a red and blue dual-color LED light source through time-sharing stroboscopic strobing to achieve layered imaging of surface and internal defects. Red light is then used to enhance the identification of reflective features of surface hair and oil stains; blue light enhances the ability to capture transmitted light, accurately locating internal defects such as splits and cracks. A white backlight source rapidly screens for highlighted defect areas through differences in light transmittance. This is combined with dual cameras for collaborative coverage, enabling multi-dimensional, high-precision detection to enhance detection effectiveness.

[0015] 2. In the present invention, mechanical vibration noise is eliminated through mean filtering and global threshold segmentation, and the accuracy of binary segmentation of defect areas is improved. Closing operation and findContours algorithm are used to effectively filter out area interference noise, retain effective defect contours, achieve image processing optimization, facilitate subsequent multimodal feature extraction, and improve detection effect.

[0016] 3. In the present invention, through the collaboration of RS485 low-latency communication, PLC real-time decision-making, and cloud-based data analysis, the glass fiber mesh cloth defect online detection system realizes a full-link closed loop from defect discovery to process optimization, improving the detection effect and quality control level. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 Flow chart of the detection method of the present invention. DETAILED DESCRIPTION

[0018] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0019] The following will be combined Figure 1 The online defect detection method for glass fiber mesh cloth based on machine vision according to an embodiment of the present invention is described in detail. Example

[0020] Reference Figure 1 , an online detection method for glass fiber mesh cloth defects based on machine vision, the detection method is: multi-channel high-speed stroboscopic light source arrangement, the red and blue dual-color LED light source is installed just above the cloth or at an angle of 15°-75° to the detection surface, and the three-dimensional layered imaging of scratches and oil stains on the fiber surface is achieved through time-sharing stroboscopic light (≥1kHz), and the white LED backlight light source is installed under the cloth to detect translucent defects such as splits and cracks. The defective area forms a highlight feature due to the transmittance; specifically, the red and blue dual-color LED light source is used to achieve layered imaging of surface and internal defects through time-sharing stroboscopic light: then the red light is used to enhance the recognition of the reflective features of surface hair balls and oil stains; the blue light improves the ability to capture transmitted light and accurately locates internal defects such as splits and cracks; the white backlight source quickly screens the highlighted defect area through the difference in transmittance, and cooperates with the dual-camera collaborative coverage, and improves the detection effect through multi-dimensional high-precision detection.

[0021] Reference Figure 1The motion platform conveyor belt is set up, and the servo motor drives the precision roller conveyor belt. The fabric is spread flat on the conveyor belt surface, with a speed stability of ±0.05% and a vibration amplitude of <5μm, ensuring strict synchronization with the camera line frequency. The encoder is installed on the conveyor belt drive shaft to provide real-time feedback of motion parameters, triggering the camera to collect and compensate for speed fluctuations. The microscope lens is set up, and a 5x microscope lens is installed on the side of the motion platform. In combination with the area array camera, secondary imaging of suspected defect areas is performed to achieve local magnification detection of subtle defects such as micron-level broken edges and white lines. Specifically, the servo motor achieves a conveyor belt speed stability of ±0.05% through closed-loop control, and the uniform tension distribution of the precision rollers prevents the fabric from wrinkling or deflecting during high-speed operation (≤50 m / min), ensuring the geometric consistency of image acquisition, effectively suppressing the interference of mechanical vibration on imaging, and avoiding image blur or misalignment. Through the coordinated design of the servo motor's high-precision transmission, encoder synchronous control, and microscope lens secondary imaging, the glass fiber mesh cloth defect detection system achieves full-area coverage and precise identification of micro defects, improving the detection effect.

[0022] Reference Figure 1 , image preprocessing, noise elimination and image segmentation of the captured image, converting the color image into a grayscale image, reducing the data dimension and unifying the feature expression, using the mean filter to smooth the captured image, eliminating the random noise caused by mechanical vibration, and when segmenting the image, applying the global threshold method to achieve binary segmentation of the fiber substrate and the defect, marking the defect area as a white pixel block, and then strengthening the high-frequency edge information, improving the continuity characteristics of linear defects such as scratches and cracks, defect detection and feature extraction, eliminating interference points with an area of ​​<0.2mm² through closing operations, retaining the effective defect contour, and using the findContours function to extract the defect geometric parameters (area ≥0.1mm², length ≥4.2 5mm, etc.) as the classification basis. When detecting bright spot defects, the mirror reflection characteristics are combined and the grayscale value sudden drop (Δ>60) is used for positioning. When detecting broken warp / weft, the Hough transform is used to detect the angular offset of the warp and weft breaks (≥15° is considered abnormal). When detecting foreign body contamination, the HSV color space is used to analyze the saturation abnormality of the color spot area (S value >120). Specifically, mean filtering and global threshold segmentation are used to eliminate mechanical vibration noise and improve the accuracy of binary segmentation of defective areas. Closing operations and findContours algorithms are used to effectively filter out area interference noise, retain effective defect contours, and achieve image processing optimization, facilitating subsequent multimodal feature extraction and improving detection effects.

[0023] Reference Figure 1, a real-time defect feedback system was built, using the RS485 communication protocol, triggering an audible and visual alarm within 50ms after a defect was detected, synchronously locating the defect coordinates, data management, and establishing an SQL database to store the defect map, position coordinates, and processing records of each roll of cloth to support data backtracking. Through the intelligent cloud platform, statistical defect distribution heat map is used to analyze process parameters and defect rates; specifically, through the collaboration of RS485 low-latency communication, PLC real-time decision-making, and cloud data analysis, the glass fiber mesh cloth defect online detection system realizes a full-link closed loop from defect discovery to process optimization, improving detection effects and quality control levels.

[0024] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or device comprising the element.

[0025] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A method for online detection of defects in glass fiber mesh cloth based on machine vision, characterized in that: The detection method is: S1: A multi-channel, high-speed stroboscopic light source arrangement uses a red and blue LED light source mounted directly above the fabric or at an angle of 15°-75° to the inspection surface. Time-sharing stroboscopic light (≥1kHz) is used to achieve three-dimensional layered imaging of scratches and oil stains on the fiber surface. A white LED backlight is mounted below the fabric to detect translucent defects such as splits and cracks. The defective area becomes a highlight due to the light transmission. S2: Camera deployment: A high-resolution line scan camera is installed perpendicular to the fabric surface, directly above the motion platform, covering a width of 1-4 meters to ensure global scanning without blind spots. Auxiliary line scan cameras are installed at an angle of 30°-45° to the side of the fabric's travel direction to enhance the ability to capture directional defects such as broken warps and wefts. S3: The motion platform conveyor belt is constructed. A servo motor drives a precision roller conveyor belt. The fabric is laid flat on the conveyor belt surface. The speed stability is ±0.05%, and the vibration amplitude is less than 5μm, ensuring strict synchronization with the camera line frequency. The encoder is installed on the conveyor belt drive shaft to provide real-time feedback of motion parameters, triggering the camera to collect and compensate for speed fluctuations. S4: Microscope lens installation: A 5x microscope lens is installed on the side of the motion platform. Combined with an area scan camera, it performs secondary imaging of suspected defect areas, enabling local magnification detection of subtle defects such as micron-level broken edges and white lines. S5: Image preprocessing: noise removal and image segmentation are performed on the captured images. Color images are converted to grayscale images to reduce data dimensionality and unify feature representation. The captured images are smoothed using a mean filter to eliminate random noise caused by mechanical vibration. During image segmentation, a global threshold method is applied to achieve binary segmentation of the fiber substrate and defects. Defective areas are marked as white pixel blocks. High-frequency edge information is then enhanced to improve the continuity characteristics of linear defects such as scratches and cracks. S6: Defect detection and feature extraction. Close operations are used to eliminate interfering noise points with an area less than 0.2 mm², retaining valid defect contours. The findContours function is used to extract defect geometric parameters (such as area ≥ 0.1 mm² and length ≥ 4.25 mm) for classification. When detecting bright spot defects, the system uses specular reflection characteristics to locate them by using a sudden drop in grayscale value (Δ> 60). When detecting broken warp and weft, the Hough transform is used to detect the angular offset of the warp and weft breaks (≥ 15° is considered an abnormality). When detecting foreign matter contamination, the HSV color space is used to analyze the saturation abnormality of the color spot area (S value > 120). S7: Real-time defect feedback system is built, using RS485 communication protocol, triggering sound and light alarms within 50ms after a defect is detected, and simultaneously locating the defect coordinates; S8: Data management: Establish an SQL database to store the defect map, location coordinates and processing records of each roll of cloth to support data backtracking, and use the intelligent cloud platform to compile statistical defect distribution heat maps to analyze process parameters and defect rates.

2. The method for online defect detection of glass fiber mesh cloth based on machine vision according to claim 1, characterized in that: The multi-channel high-speed stroboscopic light source adopts a red and blue dual-color LED light source.

3. The method for online defect detection of glass fiber mesh cloth based on machine vision according to claim 1, characterized in that: The red light channel (wavelength 630nm) in the multi-channel high-speed stroboscopic light source is used to detect surface foreign matter and hair balls, while the blue light channel (wavelength 450nm) enhances transmitted light to capture internal defects such as splits and cracks.

4. The method for online defect detection of glass fiber mesh cloth based on machine vision according to claim 1, characterized in that: The real-time defect feedback system includes a PLC controller, an audible and visual alarm, and a repeater.

5. The method for online defect detection of glass fiber mesh cloth based on machine vision according to claim 4, characterized in that: The high-resolution linear array camera captures mesh cloth images in real time, detects defects, and sends defect data to the PLC controller via RS485. The PLC controller parses the data, records the defect information, and triggers an alarm instruction. After receiving the instruction, the sound and light alarm starts the sound and light prompt to achieve real-time feedback.

6. The method for online defect detection of glass fiber mesh cloth based on machine vision according to claim 4, characterized in that: The repeater is used to dynamically adjust signal strength to ensure long-distance communication stability.

7. The method for online defect detection of glass fiber mesh cloth based on machine vision according to claim 1, characterized in that: The high-resolution line array camera uses an 8k resolution line array camera (pixel size 5μm) with a telecentric lens (FOV 1.2 meters).

Images