A foreign object detection system for rubber flooring during the vulcanization stage based on image recognition

Abstract

This invention belongs to the field of equipment monitoring, specifically disclosing an image recognition-based foreign object detection system for the vulcanization stage of rubber flooring. The system includes an imaging module, an imaging anti-interference module, and a server module. The imaging module acquires multimodal image data of the rubber flooring surface through multi-band imaging. The imaging anti-interference module isolates and processes external interference from the vulcanization environment. The server module uses image recognition technology to identify foreign objects before vulcanization and the integrity of the flooring after vulcanization, thus enabling integrated, efficient, and accurate foreign object detection during the vulcanization stage of rubber flooring.

Description

Technical Field

[0001] This invention belongs to the field of equipment monitoring and relates to a foreign object detection system for the vulcanization stage of rubber flooring based on image recognition. Background Technology

[0002] Rubber flooring is widely used in construction, medical, and sports venues due to its excellent elasticity, slip resistance, wear resistance, and environmental friendliness. In recent years, with increasing market demands for product quality, quality control in rubber flooring production has become particularly important. The vulcanization process, as a core step in rubber flooring manufacturing, directly affects the product's physical and mechanical properties. However, during vulcanization, the introduction of foreign matter (such as metal shavings and fibers) can lead to uneven vulcanization, surface defects, and even equipment damage, severely impacting product quality and production efficiency.

[0003] Currently, the foreign object detection technologies commonly used in industrial production mainly include X-ray detection and ultrasonic detection, but these technologies usually have certain limitations when applied in vulcanization workshops:

[0004] X-ray inspection solutions, such as patent 201710103634.3, utilize the penetrating power of X-rays to detect foreign objects inside adhesive strips, combined with camera-assisted surface inspection. However, X-ray equipment is expensive, and mechanical vibrations or electromagnetic noise in complex production lines may interfere with imaging accuracy and affect inspection reliability.

[0005] Ultrasonic testing solutions, such as patent 202380081495.2, utilize the emission of acoustic detection signals into and / or the reception of acoustic waves from the immersion liquid. However, ultrasonic equipment is not well-suited for use in high-temperature and high-pressure environments.

[0006] Traditional image recognition technologies used for liquid detection, such as patent 202411580987.9, utilize the characteristic that bubbles are more prominent in the backlight color channel image and foreign objects are more prominent than the background in the reflected light source color channel image. By comparing the position and region of bubbles in these two difference images, the influence of bubbles is eliminated, thereby detecting foreign objects. However, under conditions such as high temperature steam, dust, and uneven lighting, the acquired images are easily blurred or subject to severe noise interference, affecting the accuracy of foreign object identification. Summary of the Invention

[0007] In view of this, in order to solve the problems mentioned in the background technology, a foreign object detection system for the vulcanization stage of rubber flooring based on image recognition is proposed.

[0008] The objective of this invention can be achieved through the following technical solution: a foreign object detection system for the vulcanization stage of rubber flooring based on image recognition, comprising an imaging module, an imaging anti-interference module, and a server module;

[0009] The imaging module includes a multispectral imaging device, a dynamic light source adjustment device, and a 3D structured light camera. The multispectral imaging device is used to acquire images in the visible, near-infrared, and short-wave infrared bands. The dynamic light source adjustment device adjusts polarized light, coaxial light, and low-angle side light, and works in conjunction with the multispectral imaging device to dynamically switch the illumination mode according to the vulcanization process stage. The 3D structured light camera is set at the vulcanization equipment outlet to acquire three-dimensional morphological data of the rubber floor surface and detect dents or protrusions that resemble foreign objects.

[0010] The imaging anti-interference module includes an environmental isolation imaging chamber, an air curtain isolation device, and an air pressure regulating device. The environmental isolation imaging chamber is a high-temperature resistant stainless steel shell, forming a sealed imaging space inside. A high-temperature resistant optical observation window is set at the top, and a conveyor channel is set at the bottom for the continuous passage of the rubber floor to the vulcanizing equipment. Compressed air inlets and outlets are symmetrically arranged on opposite side walls along the extension direction of the conveyor channel. The air pressure regulating device is integrated on the compressed air inlets and outlets and is linked to the speed of the conveyor belt.

[0011] The server module is used to perform image recognition on the rubber floor image data acquired by the multispectral imaging device in order to distinguish foreign objects.

[0012] Preferably, the lens of the multispectral imaging device adopts an external magnetic coupling focusing mechanism, which drives the internal lens to move through a knob outside the cavity, thereby achieving focusing without opening the cavity.

[0013] Preferably, the multispectral imaging device is sealed and installed in the high-temperature resistant optical observation window via a flange interface, and the light source component of the dynamic light source adjustment device extends into the interior of the environmental isolation imaging chamber via a high-temperature resistant light guide arm.

[0014] Preferably, the air curtain isolation device is coaxially arranged around the multispectral imaging device to form a conical protective air curtain, and the spray angle of the air curtain isolation device is at an angle of 15°-30° with the optical axis of the lens of the multispectral imaging device.

[0015] Preferably, the air pressure regulating device regulates the airflow speed. With conveyor belt speed satisfy: .

[0016] Preferably, the environmental isolation imaging chamber is equipped with a temperature and humidity sensor that is linked to a dynamic light source adjustment device. When the humidity is detected to be greater than 60%RH, the illumination intensity of the short-wave infrared band is automatically increased.

[0017] Preferably, the inner surface of the environmental isolation imaging chamber is coated with a matte black light-absorbing coating with a thickness of 0.2-0.5 mm.

[0018] Preferably, the high-temperature resistant optical observation window adopts a double-layer hollow structure, with inert gas filling the middle and an anti-fog coating on the surface.

[0019] Preferably, the server module is used to perform image recognition on the rubber floor image data acquired by the multispectral imaging device to distinguish foreign objects, specifically including:

[0020] 1) Image preprocessing: The low-frequency part of the rubber floor image data is removed by a Gaussian high-pass filter, while the high-frequency part is retained to obtain a preprocessed image of the foreign object;

[0021] 2) Differential image: The preprocessed image of the foreign object is divided into odd and even components as a discrete signal. The local energy of the odd and even components is calculated by using a sliding window. An energy threshold is set to suppress noise. The denoised odd and even components are merged to reconstruct a high-frequency image.

[0022] 3) Morphological filtering detection: The high-frequency image is binarized and segmented, and opening and closing operations are performed in sequence to eliminate noise and fill holes. The area, circumscribed rectangle and centroid coordinates of each connected component are calculated, and effective foreign object regions are selected according to the preset area threshold.

[0023] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0024] (1) This invention utilizes the combination of high-frequency energy nonlinear enhancement and multi-level morphological filtering to optimize the high-temperature steam interference and complex texture background unique to the rubber vulcanization process, which greatly improves the foreign object recognition rate in the vulcanization environment.

[0025] (2) The present invention triggers image acquisition at the moment of release of sulfidation pressure or during the interval of conveyor belt, avoiding the period of maximum steam interference, and adjusts imaging parameters (such as exposure time and focal length) in real time according to sulfidation temperature and pressure to reduce the impact of environmental fluctuations.

[0026] (3) The present invention images the same area before and after vulcanization, and eliminates background interference through difference analysis. For example, cameras are deployed at multiple points such as the vulcanization bed inlet and vulcanization machine outlet to cross-verify the location of foreign objects and reduce the false detection rate. Attached Figure Description

[0027] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0028] Figure 1 This is a schematic diagram of the system structure of the present invention;

[0029] Figure 2 This is a schematic diagram of the cross-sectional structure of the system of the present invention;

[0030] Figure 3 This is a schematic diagram of the multispectral imaging device and the air curtain isolation device of the present invention;

[0031] Figure 4 This is a flowchart of the image recognition process of the imaging module of the present invention.

[0032] Explanation of the labels in the diagram:

[0033] 1-Environmental isolation imaging chamber, 2-Vulcanization equipment, 3-3D structured light camera, 4-Transfer channel, 5-Multispectral imaging device, 6-Dynamic light source adjustment device, 7-Air curtain isolation device, 8-High temperature resistant optical observation window, 9-Compressed air inlet and outlet. Detailed Implementation

[0034] 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.

[0035] This invention provides a foreign object detection system for the vulcanization stage of rubber flooring based on image recognition, including an imaging module, an imaging anti-interference module, and a server module;

[0036] like Figure 1 and Figure 2 As shown, the imaging module includes a multispectral imaging device 5, a dynamic light source adjustment device 6, and a 3D structured light camera 3. The multispectral imaging device 5 uses an InGaAs sensor (model Xenics Bobcat) and a Sony IMX490 visible light camera to achieve simultaneous imaging in the 400-2500nm wavelength band. The multispectral imaging device 5 is used to acquire images in the visible, near-infrared, and short-wave infrared bands. The dynamic light source adjustment device 6 adjusts polarized light, coaxial light, and low-angle side light, and works with the multispectral imaging device 5 to dynamically switch the illumination mode according to the vulcanization process stage. The 3D structured light camera 3 uses the LMIGOcator 3500 series, with an accuracy of 0.05mm. The 3D structured light camera is set at the vulcanization equipment outlet to acquire three-dimensional morphological data of the rubber floor surface and detect depressions or protrusions.

[0037] The imaging anti-interference module includes an environmental isolation imaging chamber 1, an air curtain isolation device 7, and an air pressure regulating device. The environmental isolation imaging chamber 1 is made of SUS304 stainless steel and has internal dimensions of 600mm×400mm×300mm. It forms a sealed imaging space inside and is equipped with a high-temperature resistant optical observation window 8 at the top. The high-temperature resistant optical observation window 8 is made of double-layer quartz glass and is filled with argon gas in the middle of the interlayer. The surface is coated with an anti-fog coating. A conveying channel 4 with a width of 500mm is set at the bottom for the continuous passage of rubber flooring to the vulcanizing equipment 2. Compressed air inlets and outlets 9 are symmetrically arranged on opposite side walls along the extension direction of the conveying channel. The air pressure regulating device is integrated on the compressed air inlets and outlets and is linked to the speed of the conveyor belt.

[0038] The server module is used to perform image recognition on the rubber floor image data acquired by the multispectral imaging device in order to distinguish foreign objects.

[0039] Preferably, the lens of the multispectral imaging device 5 adopts an external magnetic coupling focusing mechanism, which drives the internal lens to move through a knob outside the cavity, thereby achieving focusing without opening the cavity.

[0040] Preferably, the multispectral imaging device 5 is sealed and installed in the high-temperature resistant optical observation window 8 through a flange interface, and the light source component of the dynamic light source adjustment device 6 extends into the interior of the environmental isolation imaging chamber through a high-temperature resistant light guide arm.

[0041] Preferred, such as Figure 3 As shown, the air curtain isolation device 7 is coaxially arranged around the lens of the multispectral imaging device 5 to form a conical protective air curtain. The nozzle inner diameter of the air curtain isolation device 7 is 8mm, and it is arranged around the lens at a 15° tilt angle. The air is supplied after being filtered through three stages (particulate matter ≤0.1μm, oil mist ≤0.01mg / m³, dew point ≤-40℃).

[0042] Preferably, the air pressure regulating device regulates the airflow speed. With conveyor belt speed satisfy: When the system is working, the rubber floor passes through the environmental isolation imaging chamber 1 at a speed of 1 m / s, maintaining an airflow speed of 5.5 m / s.

[0043] The constant term 2.5 m / s is used to address the issue of the conveyor belt temporarily stopping. The air curtain requires a minimum velocity to maintain a basic, positive static pressure barrier. External air enters the chamber through the openings due to turbulence and pressure fluctuations, and the rising airflow from the hot air outside the imaging chamber due to density differences ensures the basic isolation effect. The proportional term... To address the dynamic interference issue, by linking the airflow speed with the conveyor belt speed and setting it slightly lower, efficient and interference-free coordination between the production process and the testing environment was achieved.

[0044] The constant term 2.5 m / s represents a basic velocity offset. Regardless of whether the conveyor belt is fast or slow, or even temporarily stopped, the air curtain requires a minimum velocity to maintain a basic, positive static pressure barrier, preventing external air from intruding through diffusion or natural convection. The proportional term... To solve the problem of dynamic interference, by linking the airflow speed with the conveyor belt speed and setting it slightly lower, efficient and interference-free coordination between the production process and the testing environment was achieved.

[0045] Specifically, the imaging chamber has a high temperature and low density inside, while the external chamber has a low temperature and high density. This density difference creates a pressure difference at the passage opening, driving cold air from the bottom to flow in and hot air from the top to flow out. Therefore, the following will occur:

[0046] ,

[0047] in, To drive pressure, The density of the external cold air. The density of the internal hot air. Gravitational acceleration, The height from the compressed air inlet and outlet to the high-temperature resistant optical observation window is [height missing]. The typical workshop temperature is 30°C. ≈1.165 kg / m³, the cabin is a high-temperature and humid environment with a temperature of 80°C. ≈ 1.000 kg / m³, gravitational acceleration ≈9.81 m / s², H is 0.5 m, therefore we can obtain ;

[0048] This pressure difference is used to accelerate air V, and its kinetic energy equation is:

[0049] ,

[0050] Will Substituting the values, we get V to be approximately 1.18 m / s. Based on engineering experience, the constant term is generally taken as 0.5-1.5 times this value, therefore the constant term is 2.5 m / s. Under similar working conditions, the drag velocity of the airflow generated by the moving surface is typically 0.5-1.0 times the main velocity, therefore the proportional term is taken as... .

[0051] Preferably, the environmental isolation imaging chamber 1 is equipped with a temperature and humidity sensor. The temperature and humidity sensor adopts SHT35 and has a measurement accuracy of ±0.2℃ and ±2%RH. It is linked with the dynamic light source adjustment device 6. When the humidity is detected to be >60%RH, the illumination intensity of the short-wave infrared band is automatically increased.

[0052] The dynamic light source adjustment device 6 is linked, specifically including:

[0053] Before vulcanization, the focus of detection is on foreign matter on the raw materials. Coaxial light + visible light / near infrared mode is used to comprehensively scan surface contaminants. During the vulcanization process, the temperature is high and may be accompanied by steam. The dynamic light source adjustment device 6 will automatically enhance the short-wave infrared illumination and use the short-wave infrared camera for imaging. Because short-wave infrared has a stronger penetrating ability to steam and high temperature, it can effectively overcome environmental interference.

[0054] For highly reflective foreign objects, such as metal shavings, the dynamic light source adjustment device 6 will use polarized light illumination, combined with the polarization filter at the imaging end, to eliminate reflections and clearly present the foreign object itself. For foreign objects whose color is similar to the rubber background, the dynamic light source adjustment device 6 will switch to monochromatic light illumination of a specific wavelength, such as near-infrared, to maximize the contrast between the foreign object and the background. At this time, the multispectral imaging device will switch to the corresponding band for acquisition.

[0055] Preferably, the lens of the multispectral imaging device 5 is an EF-S 60mm macro lens. Under the control of the server module, the knob outside the cavity drives the internal lens to move through a magnetic coupling mechanism. Each rotation of the focusing knob corresponds to a focal length adjustment of 0.5mm, with an accuracy of 0.01mm, realizing focusing without opening the cavity.

[0056] Preferably, the inner surface of the environmental isolation imaging chamber 1 is coated with a matte black light-absorbing coating, and the inner surface of the chamber is coated with a 0.3mm thick matte black paint with a reflectivity of <5%.

[0057] Preferred, such as Figure 4 As shown, the server module is used to perform image recognition on the rubber floor image data acquired by the multispectral imaging device to distinguish foreign objects, specifically including:

[0058] 1) Image preprocessing: Fourier transform (FFT) is performed on the rubber floor image data to obtain its frequency domain representation. Based on the smooth attenuation of a Gaussian high-pass filter, low-frequency signals are directly blocked while high-frequency signals are allowed to pass. Then, the image is recovered by inverse Fourier transform (IFFT) to obtain the preprocessed image of the foreign object.

[0059] 2) Differential Image: The preprocessed image of foreign objects is divided into odd and even components as a discrete signal. The even component usually contains smooth regions and low-frequency noise, while the odd component contains edges, textures and high-frequency noise. The sum of the absolute values ​​of the local windows of the odd and even components is calculated, and regions with energy above the threshold (real edges / textures) are retained, while low-energy regions (noise) are suppressed. The threshold is determined by the adaptive Otsu algorithm. The denoised odd and even sequences are merged into a high-frequency image.

[0060] 3) Morphological filtering detection: For high-frequency images, foreign objects and background are separated by adaptive threshold segmentation. Opening operation is performed sequentially, first erosion and then dilation, to remove small noise. Then closing operation is performed, first dilation and then erosion, to fill small holes. The area, circumscribed rectangle and centroid coordinates of each connected component are calculated. Based on the preset area threshold, connected components that are too small (noise) or background areas that are too large are excluded, and effective foreign object areas are selected. The number, area and location of foreign objects are recorded.

[0061] Overall, the semi-finished rubber flooring is fed into the system via conveyor channel 4 and moves towards the vulcanizing equipment 2; when the system starts, the air pressure regulating device integrated on the compressed air inlet and outlet 9 begins to work, adjusting the pressure according to the conveyor belt speed. According to the formula Calculate the required airflow velocity and control the compressed air at that velocity. Injected from the entrance and expelled from the opposite exhaust port, this airflow forms a dynamic air curtain barrier at the entrance of the transmission channel.

[0062] The rubber flooring enters the sealed imaging space inside the imaging chamber. The dynamic light source adjustment device 6 automatically selects and switches the illumination mode (such as polarized light, coaxial light, low-angle side light, and adjusts the short-wave infrared intensity) according to the preset program or feedback from the temperature and humidity sensor (such as when the humidity is >60%RH). Under the optimal illumination conditions, the multispectral imaging device 5 simultaneously acquires image data of the visible light, near-infrared and short-wave infrared bands of the rubber flooring surface. At the same time, the 3D structured light camera 3 located at the vulcanization equipment outlet scans the vulcanized flooring surface to obtain high-precision three-dimensional point cloud data.

[0063] The acquired multimodal image data is transmitted to the server module, which performs image recognition. The system can then trigger subsequent actions based on the results.

[0064] The above content is merely an example and illustration of the concept of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the concept of the invention or exceed the scope defined by the present invention, and all such modifications and additions should fall within the protection scope of the present invention.

Claims

1. A foreign object detection system for rubber flooring during the vulcanization stage based on image recognition, characterized in that, Includes an imaging module, an imaging anti-interference module, and a server module; The imaging module includes a multispectral imaging device, a dynamic light source adjustment device, and a 3D structured light camera. The multispectral imaging device is used to acquire images in the visible, near-infrared, and short-wave infrared bands. The dynamic light source adjustment device adjusts polarized light, coaxial light, and low-angle side light, and works with the multispectral imaging device to dynamically switch the illumination mode according to the vulcanization process stage. The 3D structured light camera is set at the vulcanization equipment outlet to acquire three-dimensional morphological data of the rubber floor surface, thereby detecting depressions or protrusions that resemble foreign objects. The imaging anti-interference module includes an environmental isolation imaging chamber, an air curtain isolation device, and an air pressure regulation device. The environmental isolation imaging chamber is made of high-temperature resistant stainless steel shell, forming a sealed imaging space inside. A high-temperature resistant optical observation window is set at the top, and a conveying channel is set at the bottom for the continuous passage of rubber flooring to the vulcanization equipment. Compressed air inlets and outlets are symmetrically arranged on opposite side walls along the extension direction of the conveying channel. The air pressure regulating device is integrated on the compressed air inlet and outlet and is linked to the speed of the conveying channel. The air curtain isolation device is coaxially arranged with the multispectral imaging device and surrounds its lens to form a conical protective air curtain. The spray angle of the air curtain isolation device is at an angle of 15°-30° with the optical axis of the lens of the multispectral imaging device. The air pressure regulating device adjusts the airflow speed. With the speed of the transmission channel satisfy: ; The constant term 2.5 m / s in the formula is used to establish a basic air curtain barrier to overcome the pressure difference caused by the temperature difference between the inside and outside of the imaging chamber. The resulting natural convection intrusion, proportional term Used to solve dynamic interference problems; Pressure difference The calculation formula is: , The density of the external cold air. The density of the internal hot air. Gravitational acceleration, The vertical distance from the compressed air inlet to the center of the high-temperature resistant optical observation window; The proportional term Through pressure difference The acquisition includes, specifically: Where V represents the airflow acceleration speed; The proportional term is obtained by comparing the determined airflow acceleration rate with the engineering safety factor. ; The server module is used to perform image recognition on the rubber flooring image data acquired by the multispectral imaging device to distinguish foreign objects, specifically including: 1) Image preprocessing: The low-frequency part of the rubber floor image data is removed by a Gaussian high-pass filter, while the high-frequency part is retained to obtain a preprocessed image of the foreign object; 2) Differential image: The preprocessed image of the foreign object is divided into odd and even components as a discrete signal. The local energy of the odd and even components is calculated by using a sliding window. An energy threshold is set to suppress noise. The denoised odd and even components are merged to reconstruct a high-frequency image. 3) Morphological filtering detection: The high-frequency image is binarized and segmented, and opening and closing operations are performed in sequence to eliminate noise and fill holes. The area, circumscribed rectangle and centroid coordinates of each connected component are calculated, and effective foreign object regions are selected according to the preset area threshold.

2. The foreign object detection system for the vulcanization stage of rubber flooring based on image recognition according to claim 1, characterized in that, The lens of the multispectral imaging device adopts an external magnetic coupling focusing mechanism, which drives the internal lens to move through a knob outside the cavity, so as to achieve focusing without opening the cavity.

3. The foreign object detection system for the vulcanization stage of rubber flooring based on image recognition according to claim 1, characterized in that, The multispectral imaging device is sealed and installed in the high-temperature resistant optical observation window through a flange interface, and the light source component of the dynamic light source adjustment device extends into the interior of the environmental isolation imaging chamber through a high-temperature resistant light guide arm.

4. The foreign object detection system for the vulcanization stage of rubber flooring based on image recognition according to claim 1, characterized in that, The environmental isolation imaging chamber is equipped with a temperature and humidity sensor, which is linked to a dynamic light source adjustment device. When the humidity is detected to be greater than 60%RH, the illumination intensity of the short-wave infrared band is automatically increased.

5. The foreign object detection system for the vulcanization stage of rubber flooring based on image recognition according to claim 1, characterized in that, The inner surface of the environmental isolation imaging chamber is coated with a matte black light-absorbing coating with a thickness of 0.2-0.5 mm.

6. The foreign object detection system for the vulcanization stage of rubber flooring based on image recognition according to claim 1, characterized in that, The high-temperature resistant optical observation window adopts a double-layer hollow structure, filled with inert gas in the middle, and coated with an anti-fog coating on the surface.

Images