Visual inspection apparatus and method for a combination gasket

Abstract

The application discloses a kind of visual detection device and method of combination gasket, the method includes: automatic feeding mechanism, the combination gasket to be detected as material is automatically fed to the preset detection position on rotating mechanism;Rotating mechanism, rotation is carried out;Position adjustment mechanism, the inclination angle of material on the preset detection position is adjusted;Camera mechanism, material is photographed, and the picture of the upper surface, lower surface, inner ring and outer surface of material is obtained;Control system, according to the picture obtained, the detection model obtained by pre-training is used to detect the preset defect of material, to determine whether material belongs to normal material or defective material;Discharging mechanism, if material is normal material, then material is discharged to the preset normal material discharge box, if material is defective material, then material is discharged to the preset defective material discharge box.The scheme, through fully automatic feeding and discharging, effectively detects the defects of each surface of combination gasket, reduces manual labor intensity and improves production efficiency.

Description

Technical Field

[0001] This invention belongs to the field of combined gasket inspection technology, specifically relating to a visual inspection device and method for combined gaskets, and more particularly to a rotary visual inspection automated device and method for combined gaskets. Background Technology

[0002] With the advancement of Industry 4.0, major manufacturing enterprises are setting higher standards for the quality and production capacity of gasket products to promote high-quality development through "intelligent manufacturing." During the production process, uncontrollable factors in the production and transportation environment often lead to problems such as scratches, dents, and dimensional inconsistencies in the products. Current solutions typically rely on manual inspection for gasket testing. However, manual inspection is susceptible to subjective factors, resulting in low accuracy, efficiency, and consistency, directly impacting product quality and production capacity.

[0003] The above content is only used to help understand the technical solution of the present invention and does not represent an admission that the above content is prior art. Summary of the Invention

[0004] The purpose of this invention is to provide a visual inspection device and method for composite gaskets, which solves the problem that manual inspection of composite gaskets in related solutions is prone to subjective influence, resulting in low inspection accuracy and efficiency, which affects product quality and production capacity. The invention achieves the goal of effectively detecting defects on all surfaces of composite gaskets by setting up an automated visual inspection device, and realizing fully automated loading and unloading, which helps to reduce manual labor intensity and improve production efficiency.

[0005] This invention provides a visual inspection device for a composite gasket, wherein the composite gasket includes a rubber ring and a metal ring; the visual inspection device for the composite gasket includes an automatic feeding mechanism, a rotating mechanism, a camera mechanism, a pose adjustment mechanism, a unloading mechanism, and a control system; the automatic feeding mechanism, the camera mechanism, the pose adjustment mechanism, and the unloading mechanism are respectively configured to cooperate with the rotating mechanism; the control system is used to control the automatic feeding mechanism, the rotating mechanism, the camera mechanism, the pose adjustment mechanism, and the unloading mechanism; wherein, the automatic feeding mechanism is used, under the control of the control system, to automatically feed the composite gasket to be inspected as material to a preset detection position on the rotating mechanism; the rotating mechanism is used to rotate under the control of the control system to drive the material at the preset detection position to rotate, so that the camera mechanism can photograph the material; the pose adjustment mechanism is used to... Under control, the tilt angle of the material at the preset detection position is adjusted so that the camera mechanism can capture images of the material. The camera mechanism, under the control of the control system, works in conjunction with the rotation mechanism and the pose adjustment mechanism to capture images of the material at n angles, including the top, bottom, inner circle, and outer circle of the material; n is a positive integer greater than or equal to 2. The control system, based on the images of the material at the n angles, uses a pre-trained detection model to detect preset defects in the material to determine whether the material is normal or defective. The feeding mechanism, after the control system determines whether the material is normal or defective, feeds the material into a preset normal material feeding box if it is normal, and into a preset defective material feeding box if it is defective.

[0006] In some embodiments, the camera mechanism includes: an upper camera, a lower camera, m parallel cameras, and p oblique cameras, where m and p are both positive integers greater than or equal to 2; wherein, the upper camera is used to capture an image of the upper surface of the material; the lower camera is used to capture an image of the lower surface of the material; the m parallel cameras are used to capture images of the outer cylindrical surface of the material; and the oblique cameras are used to capture images of the inner circle of the material; wherein, the rotating mechanism is rotated counterclockwise, and when the rotating mechanism rotates counterclockwise, the... The automatic feeding mechanism automatically feeds the material to a preset detection position on the rotating mechanism; the m parallel cameras capture images of the outer cylindrical surface of the material, and the upper and lower cameras capture images of the upper and lower surfaces of the material; then, the pose adjustment mechanism adjusts the pose of the material so that the p oblique cameras capture images of the inner circle of the material; then, the m parallel cameras again capture images of the outer cylindrical surface of the material, thereby obtaining images of the material from n angles: the top, bottom, inner circle, and outer circular surface.

[0007] In some embodiments, the control system uses a pre-trained detection model to detect preset defects in the material based on images of the material's top, bottom, inner ring, and outer surface from n angles, to determine whether the material is normal or defective. This includes: determining whether the top and bottom surfaces of the material have defects based on images of the material's top and bottom surfaces using the pre-trained detection model; determining whether the inner ring of the material has defects based on images of the material's inner ring; and determining whether the outer surface of the material has defects based on images of the material's outer surface using the pre-trained detection model. If it is determined that the material's top and bottom surfaces, inner ring, and outer surface are all free of defects, then the material is determined to be normal; otherwise, the material is determined to be defective.

[0008] In some embodiments, the rotation angle of the pose adjustment mechanism is determined by the detection range of the material, the diameter of the material, and the field of view of the camera lens in the camera mechanism.

[0009] In some embodiments, the automatic feeding mechanism includes: a vibratory feeder, a feeding channel, a blocking cylinder, and a blocking plate; wherein the vibratory feeder is connected to the rotating mechanism through the feeding channel, the blocking cylinder is disposed on the side of the feeding channel leading to the rotating mechanism, and the blocking plate is disposed on the side of the feeding channel leading to the rotating mechanism and located on the side of the blocking cylinder away from the vibratory feeder.

[0010] In some embodiments, the rotating mechanism includes: a metal turntable, a glass plate, and a servo motor; wherein the glass plate is disposed on the metal turntable to provide a preset detection position; and the servo motor is disposed below the metal turntable to drive the metal turntable to rotate under the control of the control system.

[0011] In some embodiments, the posture adjustment mechanism includes: a support mechanism, upper and lower cylinders, a rotary cylinder, and a gripper mechanism; wherein, when the support mechanism is vertically arranged, the upper and lower cylinders, the rotary cylinder, the gripper cylinder, and the gripper are arranged sequentially from top to bottom on the side of the support mechanism.

[0012] In some embodiments, the feeding mechanism includes: an air blowing mechanism and a feeding port; the feeding port includes: a normal material feeding port and a defective material feeding port; the corresponding feeding port is disposed between the rotating mechanism and the corresponding feeding box; the air blowing mechanism is used to blow the corresponding material to the corresponding feeding port to achieve the screening of the type of the detected material.

[0013] In some embodiments, the control system, based on images of the material from n angles (top, bottom, inner ring, and outer surface), uses a pre-trained detection model to detect preset defects in the material to determine whether the material is normal or defective. The system further includes: preprocessing the images of the material from the n angles to obtain a preprocessed image; extracting key features from the preprocessed image to obtain a key feature image; and using the pre-trained detection model to detect preset defects in the material to determine whether the material is normal or defective. In conjunction with the aforementioned device, another aspect of the present invention provides a visual inspection method for a combination gasket corresponding to a visual inspection device for combination gaskets, comprising: controlling the automatic feeding mechanism to automatically feed the combination gasket to be inspected as material to a preset detection position on the rotating mechanism; controlling the rotating mechanism to rotate so as to rotate the material at the preset detection position, thereby causing the camera mechanism to capture an image of the material; controlling the pose adjustment mechanism to adjust the tilt angle of the material at the preset detection position, thereby causing the camera mechanism to capture an image of the material; and controlling the camera mechanism to cooperate with the rotating mechanism and the pose adjustment mechanism, under the control of the control system, to inspect the material. The process involves capturing images of the material from n angles: the top, bottom, inner ring, and outer surface; where n is a positive integer greater than or equal to 2. Based on these images, a pre-trained detection model is used to detect preset defects in the material to determine whether it is a normal or defective material. The feeding mechanism is then controlled to, after determining whether the material is normal, feed it into a preset normal material feeding box; if the material is normal, feed it into a preset defective material feeding box. The process is further refined by using a pre-trained detection model to detect preset defects in the material. A pre-trained detection model detects preset defects in the material to determine whether the material is normal or defective. This includes: using images of the top and bottom of the material, the pre-trained detection model determines whether there are defects on the top and bottom surfaces; using an image of the inner circle of the material, the pre-trained detection model determines whether there are defects on the inner circle; and using an image of the outer surface of the material, the pre-trained detection model determines whether there are defects on the outer surface. If it is determined that the top and bottom surfaces, the inner circle, and the outer surface of the material are all free of defects, then the material is determined to be normal or defective. If the material is normal, then it is determined to be defective. Based on images of the material from n angles (top, bottom, inner ring, and outer surface), a pre-trained detection model is used to detect pre-defined defects in the material to determine whether it is normal or defective. This also includes: pre-processing the images of the material from n angles to obtain a pre-processed image; extracting the main features from the pre-processed image to obtain a main feature image; and using the pre-trained detection model to detect pre-defined defects in the main feature image to determine whether it is normal or defective. In conjunction with the above method, the present invention further provides a storage medium comprising a stored program, wherein, when the program is executed, the device on which the storage medium is located executes the visual inspection method for the combined gaskets corresponding to the visual inspection device for the combined gaskets described above. Therefore, the solution of this invention, by setting up an automatic feeding mechanism, a rotating mechanism, a camera mechanism, a pose adjustment mechanism, and a unloading mechanism, utilizes the automatic feeding mechanism to feed the material of the combined gasket to be tested to the detection position on the rotating mechanism. After the material reaches the detection position on the rotating mechanism, the rotating mechanism, the pose adjustment mechanism, and the camera mechanism cooperate. During the process of the rotating mechanism driving the material to rotate, and in conjunction with the pose adjustment mechanism adjusting the pose of the material, the camera mechanism captures images of the top, bottom, inner circle, and outer circular surface of the material from various angles, obtaining the top, bottom, inner circle, and outer circular surface of the material. Images are taken from various angles, including the top, bottom, inner ring, and outer surface of the material. A pre-trained detection model is used to determine whether the material is normal or defective. If the material is normal, it is fed into the normal material feeding box via a feeding mechanism; if the material is defective, it is fed into the defective material feeding box via the feeding mechanism. Thus, by setting up an automated visual inspection device for the combined gasket, defects on each surface of the combined gasket can be effectively detected, and fully automated loading and unloading can be achieved, which helps reduce manual labor intensity and improve production efficiency. Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention.

[0014] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0015] Figure 1 This is a schematic flowchart of an embodiment of the visual inspection device for the combined gaskets of the present invention; Figure 2 This is a schematic diagram of the structure of one embodiment of the rotary visual inspection machine system in the relevant scheme; Figure 3 Schematic diagrams of different types and sizes of combination gaskets; Figure 4 This is a schematic diagram of the combined gasket selected in the solution of the present invention; Figure 5 A schematic diagram showing the positional relationship between the combined gasket and the camera when a parallel camera is used to detect the outer side of the cylindrical surface of the combined gasket; Figure 6This is a schematic diagram of an embodiment of a rotary visual inspection automation device for combined gaskets according to the present invention, wherein (a) is a three-dimensional structural diagram and (b) is a top view structural diagram; Figure 7 This is a schematic diagram of the control flow of a rotary visual inspection automation device for combined gaskets according to the present invention. Figure 8 This is a schematic diagram showing the angular relationship between the combined gasket and the camera; Figure 9 This is a schematic diagram of the defects detected in the combined gaskets. From left to right, the defects are excess material, insufficient material, metal damage, and rubber ring damage. Figure 10 This is a schematic diagram of the operation flow of a rotary visual inspection automation device for combined gaskets according to the present invention. Figure 11 This is a schematic diagram of the image defect detection process of a rotary visual inspection automation device for combined gaskets according to the present invention. Figure 12 This is a schematic diagram of the rotary assembly in a rotary visual inspection automation device for combined gaskets according to the present invention. Figure 13 This is a schematic diagram of the feeding vibratory feeder in a rotary visual inspection automation device for combined gaskets according to the present invention. Figure 14 This is a schematic diagram of the pose adjustment mechanism in a rotary visual inspection automation device for combined gaskets according to the present invention. Figure 15 This is a structural diagram of the feeding port and feeding box; Figure 16 This is a structural diagram showing the combined gasket rotating to the corresponding unloading station. Figure 17 This is a schematic flowchart of an embodiment of the visual inspection method for the combined gaskets of the present invention; Figure 18 This is a flowchart illustrating an embodiment of the visual inspection method for combined gaskets of the present invention, which performs defect detection for each type of image. Figure 19 This is a schematic flowchart of an embodiment of the visual inspection method for combined gaskets of the present invention, which performs defect detection for each type of image. Figure 20 These are schematic diagrams of the images before and after preprocessing, where (a) is the image before preprocessing and (b) is the image after preprocessing. Figure 21 This is a schematic diagram of the processing flow of the ResNet50 backbone feature extraction network; Figure 22 The image after RPN processing; Figure 23 Image of rubber ring damage. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this invention, and not all of them. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0017] Considering that product inspection of gaskets is generally done manually, which is susceptible to subjective factors, leading to issues such as low accuracy, efficiency, and consistency, directly impacting product quality and production capacity. Simultaneously, manufacturing enterprises face challenges such as difficulty in recruiting and high labor costs, driving an urgent need to shift from manual inspection to intelligent visual inspection to improve product quality and production capacity, thus increasing the market demand for automated optical inspection equipment. With the growth of the intelligent manufacturing market, Zheng Motion Technology is dedicated to addressing the main challenges faced by manufacturing enterprises in various fields. For example, its motion controller solution applied to rotary visual sorting machines can quickly and effectively resolve potential quality issues in related inspection solutions, becoming a core intelligent process control tool for ensuring product quality in the manufacturing industry.

[0018] Among them, the rotary visual sorting machine is an automated inspection device developed to address defects in the size and appearance of various parts. The rotary visual sorting machine uses a vibratory feeder or other feeding mechanism to automatically transport parts to a glass turntable. Image processing is then performed on the parts on the glass turntable at the visual inspection station. The motion controller ensures constant speed rotation of the glass turntable, encoder feedback of material position, hardware position comparison output, and controls the multi-station CCD cameras and air blowing devices, achieving fully automated multi-faceted inspection and sorting of appearance, size, and other aspects.

[0019] Rotary visual sorting machines are widely used in industrial manufacturing fields such as 3C electronics, lithium batteries, electronic components, semiconductors, and hardware. Figure 2 This is a schematic diagram of the structure of one embodiment of the rotary visual inspection machine system in a related solution. For example... Figure 2 The rotary vision inspection machine system shown below operates on the following principles: (1) First, the parts to be inspected (where blue squares represent complete materials and blue triangles represent defective materials) are transported to the high-transparency glass turntable by a vibratory feeder. Then, the incoming parts are sensed by a sensor. The actual position of the incoming materials is recorded by the sensor (red squares in the figure represent sensors) and the equipment background control system.

[0020] (2) When the turntable rotates sequentially to the vision inspection station (e.g.) Figure 2 When inspecting cameras such as CCD1, CCD2, CCD3, CCD4, CCD5, CCD6, CCD7, and CCD8, the motion controller triggers the camera to take pictures through the high-speed output port, then performs image processing calculations, and finally performs automatic detection and screening according to preset detection standards.

[0021] (3) When the turntable rotates the material to the unloading station, the unloading mechanism transfers the defined material from the turntable into the corresponding complete material storage position (e.g., Figure 2 OK position) and defective material storage position (such as Figure 2 (NG position in the middle) to achieve the rejection of unacceptable materials.

[0022] However, due to the different dimensional characteristics and testing requirements of each material, most of the rotary testing equipment in related solutions is dedicated to a specific purpose and cannot be suitable for many types or sizes with a single device. Therefore, the solution of this invention is designed with a dedicated loading and unloading mechanism, a position sorting mechanism on the rotary table, and a suitable angle for the positioning camera, based on the characteristics of the combined gaskets, thereby enabling fully automated testing of the combined gaskets.

[0023] In addition, composite gaskets are integrally bonded and vulcanized from a rubber ring and a metal ring. They are sealing rings used to seal threaded and flanged connections, consisting of a metal ring and a rubber sealing gasket. Composite gaskets are suitable for piping systems using oil as the medium, providing sealing for welded, ferrule, enlarged pipe fittings, plugs, and pressure systems in mechanical devices to prevent leakage of oil, fuel, water, and chemicals. Due to their simple structure, high sealing efficiency, and low cost, they are widely used in the machinery manufacturing industry. Figure 3 These are illustrative images of different types and sizes of combination gaskets. There are many types and sizes of combination gaskets, such as... Figure 3 As shown.

[0024] Combination gaskets require inspection of multiple surfaces, including the top, bottom, inner wall, and around the metal ring. Current solutions for inspecting combination gaskets cannot effectively automate loading and unloading processes or provide comprehensive inspection from all angles.

[0025] It is evident that the proposed solution cannot perform visual inspection on combined gasket products. The main drawbacks of this solution are: First, the relevant plans allocate funds specifically for rotary visual inspection equipment, and there is currently no rotary visual inspection equipment specifically for composite gasket products. The main reason why rotary inspection equipment cannot be used for composite gasket products is that the funds are dedicated to this specific purpose. Different materials require different loading and unloading mechanisms, turntable sorting mechanisms, camera positions and angles, and visual inspection software algorithms, necessitating extensive testing, hardware optimization, and deep learning software development. Objectively speaking, no company has yet developed a fully automated rotary visual inspection system specifically for composite gaskets.

[0026] Secondly, the outer area of ​​the circle being inspected is relatively small. Since there is no space inside the turntable to place a camera, inspecting the outer surface of the circle would require an irregularly shaped mounting device for a multi-angle camera, increasing costs.

[0027] Therefore, the present invention provides a visual inspection device for composite gaskets, and particularly relates to a rotary visual inspection automation device for composite gaskets. This device is designed with a vibratory feeder for composite gasket products, a glass turntable, visual cameras mounted at various angles (top, bottom, inside, and outside), a composite gasket positioning and adjustment mechanism on the glass turntable, a pneumatic non-contact unloading mechanism, and an unloading slide mechanism conforming to the shape of the composite gasket product. This device can effectively detect defects on various surfaces of the composite gasket product and achieve fully automated unloading and loading, greatly reducing manual labor intensity and improving production efficiency.

[0028] According to embodiments of the present invention, a visual inspection device for combined gaskets is provided, such as... Figure 1 The diagram shows a flowchart of an embodiment of the method of the present invention. The combined gasket includes a rubber ring and a metal ring. Figure 4 This is a schematic diagram of the combined gasket selected in the solution of the present invention. The combined gasket selected in the solution of the present invention is a commonly used type, such as... Figure 4 As shown (left side is the actual object, right side is the image from the visual inspection camera). The visual inspection device for the combined gasket includes: an automatic feeding mechanism, a rotating mechanism, a camera mechanism, a pose adjustment mechanism, a unloading mechanism, and a control system; the automatic feeding mechanism, the camera mechanism, the pose adjustment mechanism, and the unloading mechanism are respectively configured to cooperate with the rotating mechanism; the control system is used to control the automatic feeding mechanism, the rotating mechanism, the camera mechanism, the pose adjustment mechanism, and the unloading mechanism.

[0029] The automatic feeding mechanism is used to automatically feed the combined gasket to be tested as material to the preset detection position on the rotating mechanism under the control of the control system.

[0030] The rotating mechanism is used to rotate under the control of the control system to drive the material at the preset detection position to rotate, so that the camera mechanism can capture images of the material.

[0031] The pose adjustment mechanism is used to adjust the tilt angle of the material at the preset detection position under the control of the control system, so that the camera mechanism can take pictures of the material.

[0032] The camera mechanism is used, under the control of the control system, in conjunction with the rotation mechanism and the pose adjustment mechanism, to take pictures of the material and obtain images of the material from n angles, including the top, bottom, inner circle, and outer circle; where n is a positive integer greater than or equal to 2.

[0033] The control system is used to detect preset defects in the material based on images of the material from n angles (top, bottom, inner circle, and outer circle), using a pre-trained detection model, in order to determine whether the material is normal or defective.

[0034] The feeding mechanism is used to, under the control of the control system, after the control system determines whether the material is normal or defective, feed the material into a preset normal material feeding box if the material is normal, and feed the material into a preset defective material feeding box if the material is defective.

[0035] In the proposed solution, since there is no space inside the turntable to place a camera, an irregularly shaped mounting device for a multi-angle camera is required to inspect the outer circular surface. Figure 5 This diagram illustrates the positional relationship between the combined gasket and the camera when using a parallel camera to inspect the outer side of the cylindrical surface of the combined gasket. Figure 5 As shown, the rotation center of the turntable is located below the turntable, and the combined gasket material is placed above the turntable. When inspecting the outer cylindrical surface of the combined gasket, since the camera lens is best positioned facing the side, and there is no space to place the camera inside the turntable, the present invention designs a material pose adjustment mechanism to rotate the angle of the combined gasket and take another picture for inspection, ensuring that the entire outer cylindrical surface can be detected. In addition, the turntable inspection equipment of related solutions takes pictures at an angle when inspecting the outer surface of the cylinder, which will cause some features to be obscured, increase the difficulty of the vision software, and cannot completely guarantee the inspection of the entire outer cylindrical surface.

[0036] Therefore, the present invention provides a fully automatic rotary visual inspection device for composite gasket products, which can automatically load and unload materials and perform visual inspection. At the same time, considering that there are many surfaces to be inspected for composite gaskets, a material positioning adjustment mechanism is designed on the rotary table to adjust the position of the composite gaskets on the rotary table. This allows all the inner and outer circular surfaces of the composite gaskets (or other circular materials) to be inspected, while avoiding the need for complex camera angles and reducing costs.

[0037] The present invention provides a rotary visual inspection automation device for composite gaskets. By setting up a vibratory feeder for composite gasket products, a glass turntable, visual cameras installed at various angles (top, bottom, inside, and outside), a composite gasket positioning and adjustment mechanism on the glass turntable, a pneumatic air-blowing non-contact unloading mechanism, and an unloading slide mechanism conforming to the shape of the composite gasket products, it can effectively detect defects on various surfaces of composite gasket products and achieve fully automated loading and unloading, greatly reducing manual labor intensity and improving production efficiency.

[0038] In some embodiments, the camera mechanism includes: an upper camera, a lower camera, m parallel cameras, and p oblique cameras, where m and p are both positive integers greater than or equal to 2. The upper camera is used to capture images of the upper surface of the material; the lower camera is used to capture images of the lower surface of the material; the m parallel cameras are used to capture images of the outer cylindrical surface of the material; and the oblique cameras are used to capture images of the inner ring of the material. The m parallel cameras can be arranged parallel to each other in the vertical direction. The p oblique cameras can be arranged obliquely at a predetermined angle from the vertical direction, preferably to detect the inner ring of the combined gasket.

[0039] The process involves rotating the rotating mechanism counterclockwise, causing the automatic feeding mechanism to automatically feed the material to a preset detection position on the rotating mechanism; having m parallel cameras capture images of the outer cylindrical surface of the material, and having the upper and lower cameras capture images of the upper and lower surfaces of the material; then, having the pose adjustment mechanism adjust the pose of the material so that p oblique cameras capture images of the inner circle of the material; and then, having the m parallel cameras capture images of the outer cylindrical surface of the material again, thereby obtaining images of the material from n angles: the top, bottom, inner circle, and outer circular surface.

[0040] Figure 6 This is a schematic diagram of an embodiment of an automated rotary visual inspection device for combined gaskets according to the present invention, wherein (a) is a three-dimensional structural diagram and (b) is a top view structural diagram. Figure 6 As shown in Figure (a), the rotary visual inspection automation device for combined gaskets mainly comprises: a feeding structure (such as a feeding vibratory feeder), a rotary table mechanism, upper and lower surface inspection cameras (such as an upper camera and a lower camera), a material pose adjustment mechanism, several oblique cameras, several parallel cameras, a discharge port, and a discharge bin. The feeding vibratory feeder, upper camera, lower camera, oblique cameras, parallel cameras, pose adjustment mechanism, and discharge port are all arranged on the rotary table mechanism and distributed around its periphery. The discharge bin is located below the rotary table mechanism and is arranged in conjunction with the discharge port.

[0041] like Figure 6 As shown in (b), the turntable (i.e. Figure 6 When the turntable mechanism (as shown) rotates counterclockwise, the material for the combined gasket is fed from the vibrating feeder to the designated position on the turntable. It then passes through a parallel camera to inspect the outer cylindrical surface of the combined gasket, followed by top and bottom cameras to inspect the upper and lower surfaces. A posture adjustment mechanism adjusts the posture of the combined gasket, and a tilting camera inspects the inner surface of the combined gasket. Finally, a parallel camera inspects the outer cylindrical surface of the combined gasket again, achieving comprehensive inspection of the outer cylindrical surface of the combined gasket. The software system determines whether the material is intact or defective. When the material reaches the corresponding discharge port (discharge port 1 or discharge port 2), it is pneumatically blown into the discharge channel and enters the corresponding discharge bin. The combined gasket will rotate one or more times on the turntable before being discharged; this is adjusted according to the inspection requirements.

[0042] In some embodiments, the control system uses a pre-trained detection model to detect preset defects in the material based on images of the material's top, bottom, inner ring, and outer surface from n angles, to determine whether the material is normal or defective. This includes: determining whether the top and bottom surfaces of the material have defects based on images of the material's top and bottom surfaces using the pre-trained detection model; determining whether the inner ring of the material has defects based on images of the material's inner ring; and determining whether the outer surface of the material has defects based on images of the material's outer surface using the pre-trained detection model. If it is determined that the material's top and bottom surfaces, inner ring, and outer surface are all free of defects, then the material is determined to be normal; otherwise, the material is determined to be defective.

[0043] Figure 7 This is a schematic diagram of the control flow of a rotary visual inspection automation device for combined gaskets according to the present invention. Figure 7 As shown, the specific judgment process of a rotary visual inspection automation device for combined gaskets according to the present invention includes: the material of the combined gasket enters a designated position on the rotary table (such as a glass surface), and the rotary table rotates counterclockwise. It rotates to a parallel camera to photograph and inspect the outer cylindrical surface of the material to determine if there are defects on the outer cylindrical surface; it rotates to a top and bottom camera to photograph and inspect the top and bottom surfaces of the material to determine if there are defects on the top and bottom surfaces; it rotates to a posture adjustment mechanism, which rotates the combined gasket by a designated angle; it rotates to an inclined position to inspect the inner ring of the combined gasket to determine if there are defects on the inner ring; it rotates back to a parallel camera to inspect the outer cylindrical surface of the combined gasket again to determine if there are defects on the outer cylindrical surface; after all required inspection surfaces have been inspected, it rotates to the corresponding unloading position, where a pneumatic nozzle performs non-contact unloading.

[0044] In some embodiments, the rotation angle of the pose adjustment mechanism is determined by the detection range of the material, the diameter of the material, and the field of view of the camera lens in the camera mechanism.

[0045] The rotation angle of the pose adjustment mechanism is determined by the needs of material detection, the diameter of the material, and the field of view of the camera lens. For example, if the angle of the camera lens on the outer side of the cylindrical material is 60°, and the material detection requires detection of 360° of the outer surface, the general detection method is: (1) Single camera shooting, the pose adjustment mechanism rotates the material <60° each time and takes more than 6 shots; (2) Dual camera shooting, a single shot covers 120°, the pose adjustment mechanism rotates the material <120° each time and takes more than 3 shots. There is some overlap in the shooting surface each time because the image on both sides of the camera's field of view is not as clear as the one in the middle. Generally, a partial overlap is used for shooting.

[0046] Figure 8 This is a schematic diagram showing the angular relationship between the combined gasket and the camera. (Example) Figure 8 As shown, the camera lens used in this invention can capture a 120° angle on the outer surface of a 25mm diameter combined pad. Since the material requires 360° inspection, each combined pad needs to be rotated 90° by the pose adjustment mechanism and photographed four times before being unloaded. This invention uses two parallel cameras, so each combined pad rotates twice on the turntable before being unloaded. Similarly, multiple parallel cameras and pose adjustment mechanisms can be installed to reduce the number of rotations each combined pad undergoes on the turntable, thus increasing efficiency.

[0047] Figure 9 This diagram illustrates the defects detected in the assembled gaskets. From left to right, the defects are: excess material, insufficient material, metal damage, and rubber ring damage. Many types of defects were detected, with the most numerous being: excess material, insufficient material, metal damage, and rubber ring damage. For details, please refer to [link / reference needed]. Figure 9 The example shown.

[0048] In some embodiments, the automatic feeding mechanism includes: a vibratory feeder, a feeding channel, a blocking cylinder, and a blocking plate. The vibratory feeder is connected to the rotating mechanism via the feeding channel, the blocking cylinder is located on the side of the feeding channel leading to the rotating mechanism, and the blocking plate is located on the side of the feeding channel leading to the rotating mechanism, away from the vibratory feeder.

[0049] Figure 13 This is a schematic diagram of the feeding vibratory feeder in a rotary visual inspection automation device for combined gaskets according to the present invention. The feeding vibratory feeder is as follows... Figure 13As shown, the feeding power is mainly provided by a vibratory feeder. The combined pads are arranged sequentially and enter the feeding channel. A blocking cylinder is installed at the end, which, together with the blocking plate, blocks the feeding port. The blocking cylinder, in conjunction with the movement of the vibratory feeder, can open the blocking plate and start the vibratory feeder when the rotating plate reaches the feeding position, allowing the combined pads to convey one piece of material into the designated station of the rotating plate.

[0050] In some embodiments, the rotating mechanism includes a metal turntable, a glass plate, and a servo motor. The glass plate is disposed on the metal turntable to provide a preset detection position; the servo motor is disposed below the metal turntable and drives the metal turntable to rotate under the control of the control system.

[0051] Figure 12 This is a schematic diagram of the rotary assembly in a rotary visual inspection automation device for combined gaskets according to the present invention. Figure 12 As shown, the turntable assembly includes: a metal turntable, a glass plate mounted on the metal turntable, and a servo motor positioned below the metal turntable. After the material for the combined gaskets is fed, it is placed on the glass plate. The servo motor provides rotational power to the lower part of the metal turntable, which serves as the overall frame. Glass plates are installed around the turntable, with each glass plate serving as a feeding position for the combined gaskets. The combined gaskets are precisely fed onto the glass plates by an automatic feeding mechanism. After prolonged use, when the glass plates wear down (as with all glass turntable equipment), only small pieces of glass plates can be easily replaced; the entire turntable does not need to be replaced.

[0052] In some embodiments, the posture adjustment mechanism includes: a support mechanism, upper and lower cylinders, a rotary cylinder, and a gripper mechanism; wherein, when the support mechanism is vertically arranged, the upper and lower cylinders, the rotary cylinder, the gripper cylinder, and the gripper are arranged sequentially from top to bottom on the side of the support mechanism.

[0053] Figure 14 This is a schematic diagram of the pose adjustment mechanism in a rotary visual inspection automation device for combined gaskets according to the present invention. The pose adjustment mechanism is as follows: Figure 14As shown, the device consists of a support mechanism, upper and lower cylinders (i.e., vertical cylinders), a rotary cylinder, and a gripper mechanism. When the combined pad rotates to the posture adjustment position, they work together. The upper and lower cylinders descend to the designated position, and the gripper cylinder, in conjunction with the end gripper, clamps the combined pad. The upper and lower cylinders then rise, and the rotary cylinder rotates it by a designated angle. The upper and lower cylinders descend again, the gripper cylinder releases the combined pad, and finally, the posture adjustment mechanism resets. The combined pad material on the turntable is gripped, rotated at a set angle, and then lowered. Multiple parallel and oblique cameras are used to visually inspect all internal and external surfaces. The material posture adjustment mechanism can use cylinders, electric cylinders, or springs such as tension springs and torsion springs. The turntable in this invention consists of a metal body and a small piece of high-definition visual inspection glass; alternatively, a non-metallic turntable body and glass can be used. The parallel cameras and posture adjustment mechanism work together to inspect the outer surface of other cylindrical materials.

[0054] In this invention, the orientation can be adjusted to enable omnidirectional visual inspection of the inner and outer surfaces of circular materials. When the glass panes on the turntable wear down, only the worn small glass panes can be replaced, eliminating the need to replace the entire turntable. The turntable's combined shim orientation adjustment mechanism adjusts the angle of the combined shims, allowing for the detection of all inner and outer surfaces of circular materials. The turntable utilizes segmented small-area glass panes in conjunction with a metal turntable; after a period of use, when the glass wears down, only the corresponding small glass panes need to be replaced, reducing operating costs.

[0055] like Figure 5 As shown, when a parallel camera detects the outer surface of the cylindrical surface of the composite gasket, it can only detect the surface facing the camera, while the outer surface of the cylinder facing away from the camera cannot be captured. Therefore, a pose adjustment mechanism is set up to rotate the composite gasket by an angle after capturing the outer surface facing the camera, and then capture the outer surface that could not be captured before facing away from the camera. This rotation angle is determined by the field of view of the lens. It may be necessary to rotate once to capture the entire outer surface of the cylinder, or it may be necessary to rotate multiple times to capture the entire surface.

[0056] In some embodiments, the feeding mechanism includes: an air blowing mechanism and a feeding port; the feeding port includes: a normal material feeding port and a defective material feeding port; the corresponding feeding port is disposed between the rotating mechanism and the corresponding feeding box; the air blowing mechanism is used to blow the corresponding material to the corresponding feeding port to achieve the screening of the type of the detected material.

[0057] Figure 15 This is a structural diagram of the feeding port and feeding box. (Example) Figure 15As shown, after testing, the machine is designed with two or more feeding ports and feeding bins (depending on the customer's requirements for testing several materials) to process the combined gaskets. The feeding bins have a large capacity, ensuring continuous machine operation. For example, feeding port 1 is aligned with feeding bin 1, and feeding port 2 is aligned with feeding bin 2.

[0058] Figure 10 This is a schematic diagram illustrating the operation flow of a rotary visual inspection automation device for combined gaskets according to the present invention. Figure 10 As shown, the operation flow of a rotary visual inspection automation device for combined gaskets according to the present invention includes: Step 21, Loading the product to be inspected: The vibratory feeder transports the parts to a high-transparency glass turntable, which rotates at a constant speed according to a certain set speed.

[0059] Step 22: Sensing material and recording position: When the sensor senses the material, it records the current position of the material.

[0060] Step 23, Vision Processing System: This system performs visual inspection, measurement, and defect detection on the photographed material images, and transmits OK or NG signals to the PLC. Specifically, Step 23 uses a Faster-R-CNN network to detect defects in the images. A joint training method combining RPN Loss and Faster-R-CNN Loss is employed, updating model parameters through a single backpropagation. To enhance the model's ability to detect surface defects, a feature pyramid is constructed using four different stages of feature layers from ResNet50 in the backbone feature extraction section to accommodate the detection of various defects. Verification shows that the system's detection performance meets the requirements of industrial production.

[0061] Step 24, Air-blowing screening: Based on the results fed back to the PLC by the vision system and the material position calculated by the sensor, the good / defective products are screened.

[0062] In some embodiments, the control system, based on images of the material from n angles (top, bottom, inner ring, and outer surface), uses a pre-trained detection model to detect preset defects in the material to determine whether the material is normal or defective. The system further includes: preprocessing the images of the material from the n angles to obtain a preprocessed image; extracting key features from the preprocessed image to obtain a key feature image; and using the pre-trained detection model to detect preset defects in the material to determine whether the material is normal or defective.

[0063] Figure 11This is a schematic diagram illustrating the image defect detection process of a rotary visual inspection automation device for combined gaskets according to the present invention. Figure 11 As shown, the process for detecting image defects includes: Step 31: Image preprocessing to obtain a preprocessed image. For example, median and mean filters can be used to denoise the image, resulting in a preprocessed image. See [link to documentation] for details. Figure 20 The example shown. Figure 20 The images are shown before and after preprocessing. (a) is the image before preprocessing, and (b) is the image after preprocessing, specifically the filtered image.

[0064] Step 32: Extract the backbone features of the image using a backbone feature extraction network, such as the ResNet50 backbone feature extraction network. For details on the ResNet50 backbone feature extraction network, please refer to [link to relevant documentation]. Figure 21 The example shown. Figure 21 This is a schematic diagram of the processing flow of the ResNet50 backbone feature extraction network.

[0065] Step 33: Use RPN to extract enhanced features from the backbone features of the image to obtain the enhanced features of the image. (See [link to RPN documentation]). Figure 22 The image shown shows the RPN processing effect. Figure 22 The image after RPN processing.

[0066] Step 34: Based on the enhanced features of the image, perform defect judgment and bounding box regression processing to obtain the judgment structure, i.e., the judgment result of OK or NG positions. This judgment is then complete. The final output defect image can be found in [link to image]. Figure 23 The example shown. Figure 23 Image of rubber ring damage.

[0067] Preferably, in the solution of the present invention, the image processing software with the software version number of visionmaster 4.2 can be used to process the acquired material images, which helps to ensure the speed and effect of image processing.

[0068] Figure 16 This is a structural diagram showing the assembly of the gaskets as they rotate to the corresponding unloading station. (Example:) Figure 16As shown, when the combined gasket rotates to the corresponding unloading station, high-pressure gas is blown through the air nozzle, propelling the combined gasket into the L-shaped unloading channel. Gravity causes it to fall into the unloading bin below, achieving fully automated unloading of the combined gasket products. The unloading bin has a large capacity, allowing for extended periods without cleaning and ensuring continuous equipment operation. The present invention includes one loading port, two parallel position cameras, one pair of upper and lower surface detection cameras, one posture adjustment mechanism, one tilting mechanism, and two unloading ports. However, depending on actual needs, the turntable diameter can be increased, and multiple sets of these mechanisms can be installed in combination to increase efficiency.

[0069] The technical solution of this embodiment involves setting up an automatic feeding mechanism, a rotating mechanism, a camera mechanism, a pose adjustment mechanism, and a unloading mechanism. The automatic feeding mechanism feeds the material of the combined gasket to be tested to the detection position on the rotating mechanism. After the material reaches the detection position on the rotating mechanism, the rotating mechanism, the pose adjustment mechanism, and the camera mechanism cooperate. During the rotation of the material driven by the rotating mechanism, and in conjunction with the pose adjustment mechanism adjusting the pose of the material, the camera mechanism captures images of the top, bottom, inner circle, and outer circle of the material from various angles, thus obtaining images of the top, bottom, inner circle, and outer circle of the material. Images are taken from various angles on the surface of the material, including the top, bottom, inner ring, and outer ring. A pre-trained detection model is used to detect whether the material is normal or defective. If the material is normal, it is unloaded into the normal material unloading box via a feeding mechanism; if the material is defective, it is unloaded into the defective material unloading box via the same feeding mechanism. Thus, by setting up an automated vision inspection device for the combined gasket, defects on each surface of the combined gasket can be effectively detected, and fully automated loading and unloading can be achieved, which helps reduce manual labor intensity and improve production efficiency. According to embodiments of the present invention, a visual inspection method for a combined gasket is also provided, corresponding to a visual inspection device for a combined gasket. See also Figure 17 The diagram shows a flowchart of an embodiment of the method of the present invention. The visual inspection method for the combined gasket may include steps S110 to S160.

[0070] Step S110: Control the automatic feeding mechanism to automatically feed the combined gasket to be tested as material to the preset detection position on the rotating mechanism.

[0071] Step S120: Control the rotating mechanism to rotate, thereby rotating the material at the preset detection position, so that the camera mechanism can capture images of the material.

[0072] Step S130: Control the pose adjustment mechanism to adjust the tilt angle of the material at the preset detection position so that the camera mechanism can capture images of the material.

[0073] Step S140: Control the camera mechanism to take pictures of the material under the control of the control system, in conjunction with the rotation mechanism and the pose adjustment mechanism, to obtain images of the material from n angles, including the top, bottom, inner circle, and outer circle of the material; where n is a positive integer greater than or equal to 2.

[0074] Step S150: Based on the images of the material from n angles (top, bottom, inner circle, and outer circle), a pre-trained detection model is used to detect preset defects in the material to determine whether the material is normal or defective.

[0075] In some embodiments, the specific process of detecting preset defects in the material based on images of the material's top, bottom, inner ring, and outer circular surface at n angles in step S150, and using a pre-trained detection model, to determine whether the material is normal or defective, is described in the following exemplary description.

[0076] The following is combined with Figure 18 The flowchart shown is a schematic diagram of an embodiment of the method of the present invention for defect detection of each type of image. The specific process of defect detection of each type of image in step S150 is further explained, including steps S210 to S220.

[0077] Step S210: Based on the images of the top and bottom of the material, a pre-trained detection model is used to determine whether there are defects on the top and bottom surfaces of the material; based on the image of the inner circle of the material, a pre-trained detection model is used to determine whether there are defects on the inner circle of the material; and based on the image of the outer circular surface of the material, a pre-trained detection model is used to determine whether there are defects on the outer circular surface of the material.

[0078] Step S220: If it is determined that the material has no defects on the top and bottom surfaces, no defects in the inner ring, and no defects in the outer surface, then the material is determined to be a normal material; otherwise, the material is determined to be a defective material.

[0079] Figure 10 This is a schematic diagram illustrating the operation flow of a rotary visual inspection automation device for combined gaskets according to the present invention. Figure 10 As shown, the operation flow of a rotary visual inspection automation device for combined gaskets according to the present invention includes: Step 21, Loading the product to be inspected: The vibratory feeder transports the parts to a high-transparency glass turntable, which rotates at a constant speed according to a certain set speed.

[0080] Step 22: Sensing material and recording position: When the sensor senses the material, it records the current position of the material.

[0081] Step 23, Vision Processing System: This system performs visual inspection, measurement, and defect detection on the photographed material images, and transmits OK or NG signals to the PLC. Specifically, Step 23 uses a Faster-R-CNN network to detect defects in the images. A joint training method combining RPN Loss and Faster-R-CNN Loss is employed, updating model parameters through a single backpropagation. To enhance the model's ability to detect surface defects, a feature pyramid is constructed using four different stages of feature layers from ResNet50 in the backbone feature extraction section to accommodate the detection of various defects. Verification shows that the system's detection performance meets the requirements of industrial production.

[0082] Step 24, Air-blowing screening: Based on the results fed back to the PLC by the vision system and the material position calculated by the sensor, the good / defective products are screened.

[0083] In some embodiments, step S150, which involves using a pre-trained detection model to detect preset defects in the material based on images of the material's top, bottom, inner, and outer surfaces from n angles to determine whether the material is normal or defective, further includes: The following is combined with Figure 19 The flowchart of an embodiment of the method of the present invention for defect detection of each image is shown. The specific process of defect detection of each image in step S150 is further explained, including steps S310 to S330.

[0084] Step S310: Based on the images of the material from n angles (top, bottom, inner circle, and outer circle), preprocessing is performed to obtain preprocessed images of the material.

[0085] Step S320: Extract the main features from the preprocessed image of the material to obtain the main feature image of the material.

[0086] Step S330: For the main feature image of the material, use a pre-trained detection model to detect preset defects in the material to determine whether the material is normal or defective.

[0087] Figure 11 This is a schematic diagram illustrating the image defect detection process of a rotary visual inspection automation device for combined gaskets according to the present invention. Figure 11 As shown, the process for detecting image defects includes: Step 31: Image preprocessing to obtain a preprocessed image.

[0088] Step 32: Use the backbone feature extraction network (resnet50) to extract the backbone features of the image.

[0089] Step 33: Use RPN to extract enhanced features from the backbone features of the image to obtain the enhanced features of the image.

[0090] Step 34: Based on the enhanced features of the image, perform defect judgment and border regression processing to obtain the judgment structure, i.e. the judgment result of OK or NG position. This judgment is then completed.

[0091] Step S160: Control the feeding mechanism to feed the material into a preset normal material feeding box if the material is normal material, and into a preset defective material feeding box if the material is defective material.

[0092] The present invention provides a fully automatic rotary visual inspection device for composite gasket products. By incorporating a vibratory feeder specifically designed for composite gasket products, a glass turntable, visual cameras mounted at various angles (top, bottom, inside, and outside), a composite gasket positioning and adjustment mechanism on the glass turntable, pneumatic non-contact unloading, and a unloading slide mechanism conforming to the shape of the composite gasket product, it achieves automatic loading and unloading and visual inspection. Furthermore, considering the numerous surfaces requiring inspection of composite gaskets, a material positioning and adjustment mechanism is designed on the turntable. This allows for the adjustment of the composite gasket's position on the turntable, enabling the inspection of all circular inner and outer surfaces of the composite gasket (or other circular materials), while avoiding complex camera angle placement and reducing costs.

[0093] Since the processing and functions implemented by the method in this embodiment are basically the same as the embodiments, principles and examples of the aforementioned device, any details not covered in the description of this embodiment can be found in the relevant descriptions in the aforementioned embodiments, and will not be repeated here.

[0094] The technical solution of this invention, by setting up an automatic feeding mechanism, a rotating mechanism, a camera mechanism, a posture adjustment mechanism, and a unloading mechanism, utilizes the automatic feeding mechanism to feed the material of the combined gasket to be inspected to the detection position on the rotating mechanism. After the material reaches the detection position on the rotating mechanism, the rotating mechanism, the posture adjustment mechanism, and the camera mechanism cooperate. During the rotation of the material driven by the rotating mechanism, and in conjunction with the posture adjustment mechanism adjusting the posture of the material, the camera mechanism takes pictures of the top, bottom, inner circle, and outer circular surface of the material from various angles, obtaining images of the top, bottom, inner circle, and outer circular surface of the material from various angles. Based on the images of the top, bottom, inner circle, and outer circular surface of the material from various angles, a pre-trained detection model is used for detection to determine whether the material is normal or defective. If the material is normal, it is unloaded into the normal material unloading box by the unloading mechanism; if the material is defective, it is unloaded into the defective material unloading box by the unloading mechanism. This can effectively detect defects on various surfaces of the combined gasket product and achieve fully automated loading and unloading, greatly reducing manual labor intensity.

[0095] In summary, it is readily understood by those skilled in the art that, without conflict, the aforementioned advantageous methods can be freely combined and superimposed.

[0096] The above description is merely an embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of the claims of the present invention.

Claims

1. A visual inspection device for combined gaskets, characterized in that, The combined gasket includes a rubber ring and a metal ring; the visual inspection device for the combined gasket includes an automatic feeding mechanism, a rotating mechanism, a camera mechanism, a pose adjustment mechanism, an unloading mechanism, and a control system; the automatic feeding mechanism, the camera mechanism, the pose adjustment mechanism, and the unloading mechanism are respectively configured to cooperate with the rotating mechanism; the control system is used to control the automatic feeding mechanism, the rotating mechanism, the camera mechanism, the pose adjustment mechanism, and the unloading mechanism; wherein... The automatic feeding mechanism is used to automatically feed the combined gasket to be tested as material to the preset detection position on the rotating mechanism under the control of the control system. The rotating mechanism is used to rotate under the control of the control system to drive the material on the preset detection position to rotate, so that the camera mechanism can take pictures of the material. The pose adjustment mechanism is used to adjust the tilt angle of the material at the preset detection position under the control of the control system, so that the camera mechanism can take pictures of the material. The camera mechanism is used, under the control of the control system, in conjunction with the rotation mechanism and the pose adjustment mechanism, to take pictures of the material and obtain images of the material from n angles, including the top, bottom, inner circle, and outer circle; where n is a positive integer greater than or equal to 2. The control system is used to detect preset defects in the material based on images of the material from n angles (top, bottom, inner circle, and outer circle), using a pre-trained detection model, in order to determine whether the material is normal or defective. The feeding mechanism is used to, under the control of the control system, after the control system determines whether the material is normal or defective, feed the material into a preset normal material feeding box if the material is normal, and feed the material into a preset defective material feeding box if the material is defective. The automatic feeding mechanism includes: a vibratory feeder, a feeding channel, a blocking cylinder, and a blocking plate; wherein, The vibratory feeder is connected to the rotating mechanism through the feed channel. The blocking cylinder is located on the side of the feed channel leading to the rotating mechanism. The blocking plate is located on the side of the feed channel leading to the rotating mechanism and on the side of the blocking cylinder away from the vibratory feeder. The posture adjustment mechanism includes: a support mechanism, upper and lower cylinders, a rotary cylinder, and a gripper mechanism; wherein, when the support mechanism is vertically arranged, the upper and lower cylinders, the rotary cylinder, the gripper cylinder, and the gripper are arranged sequentially from top to bottom on the side of the support mechanism. The camera mechanism includes: an upper camera, a lower camera, m parallel cameras, and p oblique cameras, where m and p are both positive integers greater than or equal to 2; wherein, The upper camera is used to capture images of the upper surface of the material; the lower camera is used to capture images of the lower surface of the material; the m parallel cameras are used to capture images of the outer cylindrical surface of the material; and the oblique camera is used to capture images of the inner circle of the material. The process involves rotating the rotating mechanism counterclockwise, causing the automatic feeding mechanism to automatically feed the material to a preset detection position on the rotating mechanism; having m parallel cameras capture images of the outer cylindrical surface of the material, and having the upper and lower cameras capture images of the upper and lower surfaces of the material; then, having the pose adjustment mechanism adjust the pose of the material so that p oblique cameras capture images of the inner circle of the material; and then, having the m parallel cameras capture images of the outer cylindrical surface of the material again, thereby obtaining images of the material from n angles: the top, bottom, inner circle, and outer circular surface.

2. The visual inspection device for the combined gasket according to claim 1, characterized in that, The control system, based on images of the material from n angles (top, bottom, inner ring, and outer surface), uses a pre-trained detection model to detect preset defects in the material to determine whether it is normal or defective material, including: Based on the top and bottom images of the material, a pre-trained detection model is used to determine whether there are defects on the top and bottom surfaces of the material; based on the image of the inner circle of the material, a pre-trained detection model is used to determine whether there are defects on the inner circle of the material; and based on the image of the outer circular surface of the material, a pre-trained detection model is used to determine whether there are defects on the outer circular surface of the material. If it is determined that the material has no defects on its top and bottom surfaces, no defects on its inner ring, and no defects on its outer surface, then the material is determined to be a normal material; otherwise, the material is determined to be a defective material.

3. The visual inspection device for the combined gasket according to claim 1, characterized in that, The rotation angle of the pose adjustment mechanism is determined by the detection range of the material, the diameter of the material, and the field of view of the camera lens in the camera mechanism.

4. The visual inspection device for the combined gasket according to any one of claims 1 to 3, characterized in that, The rotating mechanism includes: a metal turntable, a glass plate, and a servo motor; wherein, The glass plate is disposed on the metal turntable to provide a preset detection position; the servo motor is disposed below the metal turntable to drive the metal turntable to rotate under the control of the control system.

5. The visual inspection device for the combined gasket according to any one of claims 1 to 3, characterized in that, The feeding mechanism includes an air blowing mechanism and a feeding port; the feeding port includes a normal material feeding port and a defective material feeding port; the corresponding feeding port is disposed between the rotating mechanism and the corresponding feeding box; the air blowing mechanism is used to blow the corresponding material to the corresponding feeding port to achieve the screening of the type of the detected material.

6. The visual inspection device for the combined gasket according to any one of claims 1 to 3, characterized in that, The control system, based on images of the material from n angles (top, bottom, inner ring, and outer surface), uses a pre-trained detection model to detect preset defects in the material to determine whether it is normal or defective material. It also includes: Based on images of the material from n angles (top, bottom, inner circle, and outer circle), preprocessing is performed to obtain preprocessed images of the material. For the preprocessed image of the material, the main features are extracted to obtain the main feature image of the material; Based on the main feature image of the material, a pre-trained detection model is used to detect preset defects in the material to determine whether the material is normal or defective.

7. A visual inspection method for a visual inspection apparatus for a combined gasket as described in any one of claims 1 to 6, characterized in that, include: Control the automatic feeding mechanism to automatically feed the combined gasket to be tested as material to the preset detection position on the rotating mechanism; Control the rotating mechanism to rotate, thereby rotating the material at the preset detection position, so that the camera mechanism can capture images of the material; The pose adjustment mechanism is controlled to adjust the tilt angle of the material at the preset detection position, so that the camera mechanism can capture images of the material. The camera mechanism is controlled by the control system to cooperate with the rotation mechanism and the pose adjustment mechanism to take pictures of the material and obtain images of the material from n angles, including the top, bottom, inner circle, and outer circle; where n is a positive integer greater than or equal to 2. Based on images of the material from n angles (top, bottom, inner circle, and outer circle), a pre-trained detection model is used to detect preset defects in the material to determine whether the material is normal or defective. The feeding mechanism is controlled to feed the material into a preset normal material feeding box if the material is normal material and into a preset defective material feeding box if the material is defective material. in, Based on images of the material from n angles (top, bottom, inner ring, and outer surface), a pre-trained detection model is used to detect preset defects in the material to determine whether it is a normal or defective material. This includes: Based on the top and bottom images of the material, a pre-trained detection model is used to determine whether there are defects on the top and bottom surfaces of the material; based on the image of the inner circle of the material, a pre-trained detection model is used to determine whether there are defects on the inner circle of the material; and based on the image of the outer circular surface of the material, a pre-trained detection model is used to determine whether there are defects on the outer circular surface of the material. If it is determined that the material has no defects on its top and bottom surfaces, no defects on its inner ring, and no defects on its outer surface, then the material is determined to be a normal material; otherwise, the material is determined to be a defective material. Based on images of the material from n angles (top, bottom, inner ring, and outer surface), a pre-trained detection model is used to detect preset defects in the material to determine whether it is a normal or defective material. This also includes: Based on images of the material from n angles (top, bottom, inner circle, and outer circle), preprocessing is performed to obtain preprocessed images of the material. For the preprocessed image of the material, the main features are extracted to obtain the main feature image of the material; Based on the main feature image of the material, a pre-trained detection model is used to detect preset defects in the material to determine whether the material is normal or defective.

Images