Method and equipment for automatically detecting and sorting kovar rings

Abstract

The invention relates to the field of intelligent detection of kovar rings, and provides an automatic detection and sorting method and equipment for kovar rings. The method comprises the steps that kovar rings are fed to a station rotating disc through a vibration disc and a vibration feeding sliding rail, overlap removing operation and positive and negative detection operation are executed in the feeding process, and it is ensured that only one kovar ring with the upper surface facing upwards enters the station rotating disc; the Kovar ring is conveyed through rotation of the station rotary table, and the Kovar ring and the station rotary table are positioned in parallel in the tangential direction through the guide plate; in the Kovar ring transportation process, images of the Kovar ring in multiple directions are collected through multiple visual cameras; the controller receives the images of the kovar ring in multiple directions and performs preprocessing, feature extraction and defect judgment to generate a detection result; and based on the detection result, the kovar rings are sorted into qualified products, unqualified products and to-be-re-detected products through the sorting mechanism. According to the technical scheme, the ultra-narrow kovar ring can be efficiently, accurately and automatically detected, and the product quality control level is improved.

Description

Technical Field

[0001] This application relates to the field of Kovar ring intelligent detection, and in particular to an automatic detection and sorting method and device for Kovar rings. Background Technology

[0002] Kovar rings are a key material for microelectronic packaging, made of 4J29 / AgCu15 alloy. Their primary function is to achieve a hermetic seal between ceramic and metal, and they are widely used in high-reliability electronic components (e.g., aerospace, medical devices). As precision ring components, Kovar rings are characterized by their small size, thinness, and high precision. Micrometer-level invisible indentations and other defects on their surface can directly compromise the hermeticity of the package, leading to component failure. Therefore, rigorous surface quality testing of Kovar rings during the production process is essential to ensure product reliability.

[0003] Currently, the main methods for inspecting Kovar rings include manual inspection and automated inspection. Manual inspection typically relies on operators examining each Kovar ring individually under a microscope. While flexible, this method is inefficient and highly susceptible to operator experience and fatigue, easily leading to missed or false detections. To address the drawbacks of manual inspection, automated inspection and sorting equipment has been developed. These devices generally utilize vibratory feeders, conveyor mechanisms, vision systems for imaging, and simple image processing algorithms to automate the inspection and sorting of Kovar rings of standard sizes. This type of equipment can improve inspection efficiency to a certain extent and reduce human error.

[0004] However, existing automated inspection and sorting equipment has significant shortcomings when dealing with ultra-narrow Kovar rings with an inner frame width of less than 1.2 mm. On the one hand, the ultra-narrow size makes it extremely easy for Kovar rings to overlap or become misaligned during vibratory feeding, and the orientation and screening mechanisms of conventional equipment are unable to effectively handle these problems, causing feeding jams or defective products entering the inspection station. On the other hand, the effective feature area provided by the ultra-narrow frame during image acquisition is extremely small, and conventional visual lighting schemes and image processing algorithms are unable to stably extract contours and surface features, resulting in poor image quality, low defect recognition rate, frequent misjudgments or inability to determine defects by the equipment, and ultimately still requiring manual re-inspection, failing to truly achieve efficient automated inspection of ultra-narrow Kovar rings. Summary of the Invention

[0005] This application provides an automatic detection and sorting method and device for Kovar rings. Through multi-camera collaborative image acquisition and intelligent judgment and sorting, it realizes efficient and accurate automated detection of ultra-narrow size Kovar rings and improves the level of product quality control.

[0006] The first aspect of this application provides an automatic detection and sorting method for Kovar rings, the method comprising: Kovar rings are fed to the station turntable using a vibratory feeder and a vibratory feeding slide rail. During the feeding process, an anti-overlap operation and a front-back detection operation are performed to ensure that only a single Kovar ring with its upper surface facing up enters the station turntable. The Kovar ring is transported by rotating on a turntable at the workstation, and the Kovar ring is positioned parallel to the tangential direction of the turntable using a guide plate. During transportation, images of the Kovar ring from multiple directions are captured using multiple vision cameras; The controller receives the image and performs preprocessing, feature extraction, and defect determination to generate detection results. Based on the test results, the Kovar ring is sorted into qualified products, unqualified products, and products to be re-inspected by a sorting mechanism.

[0007] A second aspect of this application provides an automatic Kovar ring detection and sorting device, the device comprising: Machine tool; The workstation turntable is located at the center of the machine and is connected to the rotating motor inside the machine for transporting Kovar rings; A vibratory feeder, mounted on the machine platform, is used for storing and feeding the Kovar rings; A vibratory feeding slide rail connects the discharge port of the vibratory plate and the station turntable, and is used to guide the Kovar ring; The material arrival sensor is located above the station turntable and is used to detect the arrival of the Kovar ring; Multiple vision cameras are fixedly installed along the rotation direction of the workstation turntable to acquire Kovar ring images; A sorting mechanism used to sort Kovar rings based on test results; The controller, located inside the machine, is electrically connected to the incoming material sensor, the multiple vision cameras, and the sorting mechanism, and is used to control the detection and sorting process.

[0008] The technical solution provided in this application can include the following beneficial results: On the one hand, Kovar rings are fed to the station turntable via a vibratory feeder and a vibratory feeding slide rail, and anti-overlap and forward / backward detection operations are performed during the feeding process. This automatically removes overlapping parts and Kovar rings with incorrect postures before the material enters the detection stage, ensuring that only a single product with the detection surface (top surface) facing upward enters the station turntable. This avoids detection interruptions or misjudgments caused by unstable material conditions from the source, thereby improving the automation efficiency of the entire system and the reliability of the initial stage. On the other hand, multiple vision cameras are used to acquire images of Kovar rings from multiple directions. This multi-angle image acquisition method overcomes the limitations of a single camera perspective. The limitations of traditional methods are mitigated by ensuring consistent shooting pose for each Kovar ring, effectively reducing blind spots and allowing defects in all areas of the ring surface to be captured, thus improving the comprehensiveness and consistency of the inspection. Thirdly, the controller's dedicated algorithm performs deep processing on the acquired images, enabling targeted algorithm optimization to address the weak image features and noise sensitivity of ultra-narrow-sized Kovar rings. This allows for stable extraction of real defect features under complex imaging conditions, improving the accuracy and anti-interference capabilities of the judgment results. The direct linkage between the detection results and sorting execution forms a complete automated process from detection to classification, improving overall processing efficiency and ensuring the orderly sorting of products. In summary, the technical solution of this application achieves efficient and accurate automated detection of ultra-narrow-sized Kovar rings and improves product quality control.

[0009] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description

[0010] The above and other objects, features and advantages of this application will become more apparent from the more detailed description of exemplary embodiments thereof in conjunction with the accompanying drawings, wherein the same reference numerals generally represent the same components in the exemplary embodiments thereof.

[0011] Figure 1 This is a flowchart illustrating the automatic detection and sorting method for Kovar rings as shown in the embodiments of this application; Figure 2 This is the Kovar ring automatic detection and sorting device shown in the embodiments of this application; Figure 3 This is an internal isometric view of the Kovar ring automatic detection and sorting device shown in the embodiments of this application; Figure 4 This is an external view of the Kovar ring automatic detection and sorting device shown in the embodiments of this application. Detailed Implementation

[0012] Embodiments of this application will now be described in more detail with reference to the accompanying drawings. While embodiments of this application are shown in the drawings, it should be understood that this application may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to make this application more thorough and complete, and to fully convey the scope of this application to those skilled in the art.

[0013] The terminology used in this application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. The singular forms “a,” “the,” and “the” used in this application and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any or all possible combinations of one or more of the associated listed items.

[0014] It should be understood that although the terms "first," "second," "third," etc., may be used in this application to describe various information, this information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, without departing from the scope of this application, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0015] Currently, the main methods for inspecting Kovar rings include manual inspection and automated inspection. Manual inspection typically relies on operators examining each Kovar ring individually under a microscope. While flexible, this method is inefficient and highly susceptible to operator experience and fatigue, leading to missed or false detections. To address the drawbacks of manual inspection, automated inspection and sorting equipment has been developed. These devices generally utilize vibratory feeders, conveyor mechanisms, vision systems for imaging, and simple image processing algorithms to automate the inspection and sorting of Kovar rings of standard sizes (e.g., inner ring width ≥ 1.2 mm). This type of equipment can improve inspection efficiency to some extent and reduce human error. However, existing automated inspection and sorting equipment has significant limitations when dealing with ultra-narrow Kovar rings with inner frame widths less than 1.2 mm (e.g., widths between 0.8 mm and 1.2 mm). On the one hand, the ultra-narrow size makes Kovar rings prone to overlapping or misalignment (e.g., back facing up) during vibratory feeding. Conventional equipment's orientation and screening mechanisms (e.g., simple tracks or single sensors) struggle to effectively handle these issues, causing feeding delays or defective products entering the inspection station. On the other hand, the ultra-narrow bezel provides a very small effective feature area during image acquisition. Conventional visual lighting schemes and image processing algorithms (e.g., basic threshold segmentation) struggle to stably extract contours and surface features, resulting in poor image quality, low defect recognition rates, frequent misjudgments or inability to determine defects, ultimately requiring manual re-inspection and failing to achieve truly efficient automated inspection of ultra-narrow Kovar rings.

[0016] To address the aforementioned problems in the prior art, this application proposes an automatic detection and sorting method for Kovar rings, which can be applied to... Figure 3 and Figure 4 The flowchart of the Kovar ring automatic detection and sorting equipment and the Kovar ring automatic detection and sorting method is attached. Figure 1 As shown, the main steps include S101 to S105, which are detailed below: Step S101: The Kovar ring is fed to the station turntable 2 via the vibratory feeder 3 and the vibratory feeding slide rail 4. During the feeding process, the overlapping operation and the forward and reverse detection operation are performed to ensure that only a single Kovar ring with the upper surface facing up enters the station turntable 2.

[0017] Step S101 is the initial stage of the detection process, and its core is to solve the problem of easy overlap and incorrect posture of ultra-narrow Kovar rings during the feeding process. As an embodiment of this application, the anti-overlap operation and forward / reverse detection operation during the feeding process can be performed as follows: multiple sets of anti-overlap blocks are set on the spiral track inside the vibratory feeder 3, so that the overlapping Kovar rings slide back to the bottom of the vibratory feeder 3 under the action of gravity via the sliding guide block; the reflectivity of the Kovar ring surface is detected by a reflective fiber optic probe, and when the back of the Kovar ring is facing upward, the jet controller 12 controls the jet duct to spray gas, blowing the Kovar ring back to the bottom of the vibratory feeder 3. Specifically, the vibratory feeder 3 is set on the right side of the machine base 1 and is vertically higher than the station turntable 2. Its inner wall is provided with a spiral track that rises upward, and multiple sets of anti-overlap blocks are installed on the track. The anti-overlap blocks are slightly raised above the inner wall of the vibratory feeder 3 (for example, the thickness of the protrusion can be adjusted according to the size of the Kovar ring), and a sliding guide block is set at the bottom. When the Kovar ring moves upward along the spiral track under the vibration of the vibratory feeder 3, if multiple products overlap, the overlapping parts, unable to fully conform to the narrowed inner wall of the track, will slide back to the bottom of the vibratory feeder 3 under gravity via the guide block for reloading. This de-overlapping operation effectively avoids material jamming and the inflow of defective products, improving loading reliability. Simultaneously, a Kovar ring forward / reverse detection assembly is mounted above the de-overlapping block, including a reflective fiber optic probe, an air duct, a control valve, and an air controller 12. The reflective fiber optic probe is located at the bottom of the air duct, which is connected to the air source via the control valve. The forward / reverse detection operation is based on the difference in reflectivity of the Kovar ring surface materials: the upper surface of the Kovar ring is coated with a silver-copper alloy, which has high reflectivity, while the back surface is made of Kovar alloy, which has low reflectivity. When a product passes the inspection point, the reflective fiber optic probe emits light: if the top surface faces upward, the high reflectivity causes the light to reflect back to the probe, the jet controller 12 closes the control valve, and the product passes normally; if the back surface faces upward, the low reflectivity results in no reflected signal, the jet controller 12 opens the control valve, and the jet duct sprays gas (the gas pressure is adjustable, for example, increased to 0.5~0.8MPa to handle ultra-narrow dimensions), blowing the product towards the sliding guide block and back to the bottom of the vibratory feeder. This design ensures that only a single product with the inspection surface facing upward enters the subsequent station, improving inspection accuracy from the source. Furthermore, the discharge port of the vibratory feeder 3 is fixedly connected to the inlet of the vibratory feeding slide rail 4, and the discharge port of the slide rail 4 is located above the station turntable 2. An incoming material sensor 5 (e.g., an infrared sensor) is installed above the station turntable 2 to detect the arrival of the Kovar ring. When the incoming material sensor 5 detects the Kovar ring, the controller 12 calculates the time it takes for the Kovar ring to reach each vision camera based on the rotation speed and distance of the station turntable 2, and controls the vision camera to delay shooting, thereby synchronizing image acquisition and material flow.

[0018] Step S102: The Kovar ring is transported by rotating the station turntable 2, and the guide plate is used to position the Kovar ring parallel to the tangential direction of the station turntable 2.

[0019] In this embodiment, the station turntable 2 is located at the center of the machine base 1 and is connected to a rotating motor inside the machine base to transport Kovar rings at a constant speed (e.g., 10-15 rpm). Crucially, a guide plate is fixedly installed on the machine base 1, located in front of the discharge port of the vibrating feeding rail 4. When the Kovar ring enters the station turntable 2, under the influence of inertia and the rotation of the turntable, the product comes into frictional contact with the guide plate, forcing the Kovar ring to adjust to be parallel to the tangential direction of the station turntable 2. This positioning operation ensures that each Kovar ring has a consistent pose during subsequent image acquisition, reducing blind spots and laying the foundation for multi-angle imaging. If the Kovar ring is oriented incorrectly (e.g., tilted), it cannot be correctly detected, thus avoiding misjudgment. The guide plate design is simple and reliable, requiring no additional power, thus reducing equipment complexity.

[0020] Step S103: During the transportation of the Kovar ring, images of the Kovar ring from multiple directions are acquired using multiple vision cameras.

[0021] like Figure 3 As shown, multiple vision cameras include a first vision camera 7, a second vision camera 8, a third vision camera 9, and a fourth vision camera 10. During the transport of the Kovar ring, these four vision cameras can respectively capture images of the top, bottom, left, and right sides of the Kovar ring. Specifically, the four vision cameras are fixedly set along the rotation direction of the station turntable 2: the first vision camera 7 is located at the first position after the loading station, capturing the image of the top of the Kovar ring; the second vision camera 8 captures the image of the bottom; the third vision camera 9 captures the image of the left side; and the fourth vision camera 10 captures the image of the right side. Each vision camera consists of an adjustable bracket, a vision camera, and a ring aperture. The ring aperture is located below the camera to provide a uniform light source. The adjustable bracket allows adjustment of the camera height and angle to accommodate Kovar rings of different sizes, ensuring image acquisition quality. The light source of the ring aperture (e.g., LED white light) enhances the contrast between the Kovar ring and the background, facilitating subsequent binarization processing. The image acquisition process is synchronously controlled by controller 12: after the incoming material sensor 5 detects the Kovar ring, controller 12 calculates the delay time based on the turntable rotation speed and camera spacing, and triggers each camera to capture images sequentially. This multi-angle acquisition method overcomes the drawbacks of the limited field of view of a single camera. Combined with the pose positioning in step S102, it ensures that defects such as hidden pressure marks on all four surfaces of the Kovar ring, such as the top, bottom, left, and right, can be captured, improving the comprehensiveness of the detection. For ultra-narrow Kovar rings (i.e., Kovar rings with an inner ring width ≥ 0.8 mm and < 1.2 mm), small feature areas are easily affected by noise, but multi-view images provide redundant data for subsequent processing, improving robustness.

[0022] Step S104: The controller 12 receives images of the Kovar ring from multiple directions captured by multiple vision cameras and performs preprocessing, feature extraction and defect determination to generate detection results.

[0023] The controller 12 receives images of Kovar rings from multiple vision cameras in multiple directions and performs depth processing to identify defects in ultra-narrow Kovar rings, such as hidden pressure marks, which is the core step in the inspection process.

[0024] The processing flow of step S104 includes three stages: preprocessing, feature extraction, and defect determination. It combines specialized algorithms to improve the recognition accuracy of ultra-narrow border images. The preprocessing stage first optimizes the original image to enhance effective features and suppress noise. Optionally, preprocessing includes the following steps: median filtering algorithm, linear grayscale stretching algorithm, bimodal histogram thresholding method, image morphological processing, and connected component analysis. These steps are described in detail below: 1) Median filtering algorithm: For each pixel in the image acquired by the vision camera, the pixel values ​​in its surrounding area (e.g., a 3×3 or 5×5 window) are sorted, and the median value is assigned to that pixel. This algorithm can effectively remove salt-and-pepper noise and impulse noise while preserving edge details and reducing noise interference to subsequent processing.

[0025] 2) Linear gray-scale stretching algorithm: This algorithm expands the dynamic range of image gray levels through mapping relationships, enhancing the contrast between the Kovar ring and the background. Specifically, it linearly transforms the original gray values ​​to the full range (e.g., 0~255), making the subtle features of the ultra-narrow bezel more prominent.

[0026] 3) Bimodal histogram thresholding method: This method analyzes the peak and valley values ​​of the image histogram to find the optimal threshold to form a clear bimodal structure, thereby extracting the binarized image of the Kovar ring. Controller 12 presets specific low and high thresholds for each size of Kovar ring to ensure segmentation adaptability.

[0027] 4) Image morphological processing: First, perform erosion to remove small white noise, then perform dilation to enlarge the foreground object and reduce noise in the binarized image. Optionally, use grayscale morphological operations of the VisionPro CogBlob Tool, such as square erosion to eliminate small noise, square dilation to fill edge gaps, and closing operations to eliminate surface holes.

[0028] 5) Connected component analysis processing, namely: marking the regions composed of adjacent pixels in the binarized image, filtering background interference spots based on the preset minimum area parameter, and extracting the complete connected region of Kovar ring.

[0029] The feature extraction and defect determination stages in the above embodiments are based on the preprocessed image. Optionally, the controller 12 uses the VisionPro blob tool to extract feature parameters such as the area, centroid coordinates, and edge integrity ratio of the Kovar ring, and compares these feature parameters with a standard parameter library (which can be established based on qualified Kovar ring samples). If the feature difference exceeds a set threshold (e.g., area deviation > 5% or edge integrity < 95%), it is determined to be a defective product (i.e., there is hidden pressure damage); if the difference is within the threshold, it is a qualified product; if the image quality is abnormal (e.g., overexposed or blurred), it is marked as a product to be re-inspected. To further optimize the detection robustness of ultra-narrow dimensions, the controller 12 also uses advanced algorithms such as adaptive gradient weighted median filtering algorithm and multi-scale contour fusion analysis, wherein: The adaptive gradient-weighted median filtering algorithm specifically calculates the gradient value of each pixel in the image and dynamically adjusts the size and weight of the median filtering window. For edge regions with high gradient values ​​(e.g., the contour of Kovar rings), a small window (e.g., 3×3) is used to preserve details; for flat regions with low gradient values, a large window (e.g., 7×7) is used to suppress noise. This algorithm balances detail preservation and noise reduction, addressing the characteristics of weak features and noise sensitivity in ultra-narrow borders. Multi-scale contour fusion analysis, on the other hand, performs edge detection at different scales on the binarized image (e.g., Canny operator multi-threshold processing), extracting multi-level contour features from coarse to fine. Then, based on the continuity of contour curvature, it fuses these features to reconstruct a highly complete edge contour. This algorithm effectively solves the problem of contour breakage caused by imaging blur in ultra-narrow borders.

[0030] As can be seen from the above embodiments, through a combination of specialized algorithms, the Kovar ring automatic detection and sorting equipment can stably extract the true defect features of ultra-narrow Kovar rings under complex imaging conditions, avoiding misjudgment. For example, grayscale stretching and threshold segmentation improve contrast, while morphological processing ensures edge continuity, ultimately achieving a sorting accuracy of >85%.

[0031] Step S105: Based on the test results, Kovar rings are sorted into qualified products, unqualified products, and products to be re-inspected by the sorting mechanism 11.

[0032] Steps S104 and S105 in the above embodiment demonstrate the direct linkage between the sorting process and the detection results, achieving closed-loop automation. Specifically, the sorting mechanism 11 includes three sets of spray guns spaced apart and three sets of discharge collection boxes, corresponding to the collection of qualified products, unqualified products, and products awaiting re-inspection, respectively. The controller 12 calculates the delay time based on the rotation speed of the station turntable 2 and the distance between the fourth vision camera 10 and the sorting mechanism 11, and controls the corresponding spray gun to start: when the Kovar rotates to the sorting station, the spray gun sprays gas to blow the product into the corresponding collection box. For example, if the detection result is unqualified, the unqualified product spray gun starts, and the product falls into the unqualified box. This design ensures the timeliness and accuracy of the sorting action, with a detection speed of 190-230 pieces / minute. To handle uncertainties, Figure 1 The example method can also include closed-loop feedback re-inspection, whereby, for Kovar rings determined to require re-inspection, the controller 12 redirects them back to the vibratory feeder 3 for loading, and upon passing the vision camera again, performs a fusion analysis (e.g., weighted averaging or feature complementarity) combining the current image with the previous image, ultimately providing a deterministic sorting result. This mechanism reduces misjudgments caused by temporary interference (e.g., light fluctuations), improving system reliability. Furthermore, Figure 1 The example method also introduces a dynamic parameter optimization mechanism to adapt to production fluctuations. Specifically, during continuous inspection, the controller 12 calculates the recent sorting pass rate in real time. If the pass rate consistently deviates from the preset target (e.g., 85%~90%), the controller 12 automatically fine-tunes the high / low threshold combination in the bimodal histogram threshold segmentation method (e.g., the low / high threshold combination in the bimodal threshold segmentation method) to bring the sorting pass rate back to the target range, achieving adaptive optimization of the inspection parameters. Optionally, the controller 12 employs a reinforcement learning-based decision model to select the threshold adjustment strategy. The model takes the pass rate deviation and historical adjustment effects as state inputs and the threshold adjustment direction and magnitude as action outputs, learning the optimal parameter adjustment strategy through continuous interaction with the environment. For example, if the pass rate is too low, the decision model may appropriately lower the grayscale screening threshold to capture more subtle defects. This adaptive mechanism enables the equipment to maintain stable performance over a long period.

[0033] From the above appendix Figure 1The example of the Kovar ring automatic detection and sorting method demonstrates that, on the one hand, Kovar rings are fed to the station turntable via a vibratory feeder and vibratory feeding rail. During the feeding process, anti-overlap and forward / reverse detection operations are performed. This automatically removes overlapping parts and Kovar rings with incorrect orientation before the material enters the detection stage, ensuring that only a single product with the detection surface (top surface) facing upward enters the station turntable. This avoids detection interruptions or misjudgments caused by unstable material conditions from the source, thereby improving the automation efficiency of the entire system and the reliability of the initial stage. On the other hand, multiple vision cameras acquire images of Kovar rings from multiple directions. This multi-angle image acquisition method overcomes the limited field of view of a single camera. To address the drawbacks of traditional methods, this application ensures that each Kovar ring maintains a consistent shooting pose, effectively reducing blind spots and allowing defects in all areas of the ring surface to be captured, thus improving the comprehensiveness and consistency of the inspection. Thirdly, the controller's dedicated algorithm performs deep processing on the acquired images, enabling targeted algorithm optimization to address the weak image features and noise sensitivity of ultra-narrow-sized Kovar rings. This allows for stable extraction of real defect features under complex imaging conditions, improving the accuracy and anti-interference capabilities of the judgment results. The direct linkage between the detection results and sorting execution forms a complete automated process from detection to classification, improving overall processing efficiency and ensuring the orderly sorting of products. In summary, the technical solution of this application achieves efficient and accurate automated detection of ultra-narrow-sized Kovar rings and improves product quality control.

[0034] Corresponding to the aforementioned application function implementation method embodiments, this application also provides an embodiment of an automatic Kovar ring detection and sorting device.

[0035] Figure 2 This is a schematic diagram of an automatic Kovar ring detection and sorting device according to an embodiment of this application. The device may include a machine base 1, a turntable 2, a vibratory feeder 3, a vibratory feeding rail 4, a material receiving sensor 5, multiple vision cameras, a sorting mechanism 11, and a controller 12, as detailed below: The workstation turntable 2 is located at the center of the machine base 1 and is connected to the rotating motor inside the machine base 1 for transporting Kovar rings; Vibratory feeder 3, installed on machine base 1, is used for storing and feeding Kovar rings; Vibratory feeding slide rail 4 connects the discharge port of vibratory plate 3 and station turntable 2, and is used to guide Kovar ring; The incoming material sensor 5 is located above the station turntable 2 and is used to detect the arrival of the Kovar ring; Multiple vision cameras are fixedly set along the rotation direction of the workstation turntable 2 to acquire Kovar ring images; Sorting mechanism 11 is used to sort Kovar rings based on test results; The controller 12 is located inside the machine 1 and is electrically connected to the incoming material sensor 5, multiple vision cameras and sorting mechanism 11 to control the detection and sorting process.

[0036] See Figure 3 and Figure 4 Its core lies in integrating mechanical structures, sensing systems, and control units to address the challenges of feeding, inspecting, and sorting ultra-narrow Kovar rings. The equipment comprises the following components, which are electrically and mechanically connected to achieve closed-loop automation: The machine base 1, serving as the support frame for the equipment, is made of rigid materials (such as aluminum alloy) and houses a rotating motor and controller 12, providing structural stability and a power foundation. The surface of the machine base 1 has mounting holes for installing other components, ensuring a compact overall layout and facilitating industrial deployment.

[0037] The turntable 2 is located at the center of the machine base 1 and is connected to the rotating motor inside the machine base via a shaft system. The turntable rotates at an adjustable speed (e.g., 10~15 rpm) for continuous transport of Kovar rings. The edge of the turntable is provided with evenly distributed positioning grooves to temporarily fix the Kovar rings and prevent them from slipping or shifting during transport.

[0038] Vibratory feeder 3 is mounted on the machine base 1, typically located on the right side and vertically higher than the station turntable 2, for storing and initially feeding Kovar rings. The inner wall of the vibratory feeder has a spiral track that serves as the feeding channel for the Kovar rings. Crucially, multiple sets of anti-overlap blocks are installed on the track. These anti-overlap blocks protrude from the inner wall of the vibratory feeder (the thickness of the protrusion can be finely adjusted according to the Kovar ring size, for example, 0.5~1mm), and a sliding guide block is located at the bottom. When the Kovar rings rise along the track under vibration, if the products overlap, the overlapping portion loses support due to the narrowing of the track and will slide back to the bottom of the vibratory feeder under gravity via the sliding guide block, achieving automatic anti-overlap. This mechanical design is simple and reliable, avoiding material jamming problems.

[0039] The vibratory feeding slide rail 4 connects the discharge port of the vibratory plate 3 and the station turntable 2. It adopts a guide rail structure to guide the Kovar ring to the station turntable 2. The tilt angle and vibration frequency of the slide rail are adjustable to ensure smooth conveying of the Kovar ring and reduce collisions or posture changes.

[0040] The incoming material sensor 5 is positioned above the station turntable 2, and is an infrared photoelectric sensor used to detect the arrival of the Kovar ring. When a product is detected, the incoming material sensor 5 sends a signal to the controller 12. The controller calculates the time delay based on the rotation speed of the station turntable 2 and the camera spacing to synchronize subsequent image acquisition.

[0041] The vision camera system includes a first vision camera 7, a second vision camera 8, a third vision camera 9, and a fourth vision camera 10, fixedly positioned along the rotation direction of the workstation turntable 2. Each vision camera consists of an adjustable bracket, a vision camera, and a ring aperture: the adjustable bracket is used to adjust the height and angle of the camera to accommodate Kovar rings of different sizes; the ring aperture is located below the camera, providing uniform LED light source, enhancing the contrast of image acquisition, and facilitating subsequent binarization processing. The cameras acquire images of the top, bottom, left, and right sides of the Kovar ring, ensuring multi-angle coverage and reducing blind spots.

[0042] The sorting mechanism 11 sorts Kovar rings based on the test results, and includes three sets of spray guns spaced apart and three sets of discharge collection boxes. The spray guns are connected to an air source and are controlled by a controller 12 to start with a delay. The collection boxes correspond to qualified products, unqualified products, and products to be re-inspected, respectively, to achieve orderly classification.

[0043] The controller 12 is located inside the machine 1 and serves as the control core. It is electrically connected to the incoming material sensor 5, the vision camera, and the sorting mechanism 11. The controller is configured to preset detection parameters for each size of Kovar ring, such as the low and high thresholds of the bimodal histogram threshold segmentation method, the minimum area of ​​connected component analysis, and the grayscale screening threshold, to ensure algorithm adaptability and detection consistency.

[0044] From the above appendix Figure 2As illustrated by the example of the Kovar ring automatic detection and sorting equipment, on the one hand, the vibratory feeder, vibratory feeding rail, and the anti-overlapping block and forward / reverse detection components integrated into the vibratory feeder automatically remove overlapping parts and Kovar rings with incorrect orientation before the material enters the detection stage. This ensures that only a single product with its detection surface (top surface) facing upward enters the station turntable, avoiding detection interruptions or misjudgments caused by unstable material conditions from the source, thus significantly improving the automation efficiency and reliability of the entire system's initial operation. On the other hand, multiple vision cameras (i.e., the first vision camera, the second vision camera, the third vision camera, and the fourth vision camera) fixedly set along the rotation direction of the station turntable can simultaneously or sequentially acquire complete surface images of the Kovar rings from four directions: up, down, left, and right. This multi-angle image acquisition method overcomes the drawbacks of the limited field of view of a single camera, combined with the advantages of... The guide plate positions the Kovar ring parallel to the tangent of the rotary table, ensuring consistent shooting posture for each Kovar ring. This effectively reduces blind spots and allows defects in all areas of the ring surface to be captured, improving the comprehensiveness and consistency of the inspection. Thirdly, the electrical signal connection structure between the controller and each vision camera enables targeted algorithm optimization for the weak image features and noise sensitivity of ultra-narrow-sized Kovar rings. This allows for stable extraction of real defect features under complex imaging conditions, improving the accuracy and anti-interference capability of the judgment results. Fourthly, the linkage between the sorting mechanism and the controller allows for immediate driving of the sorting action after quality judgment, ensuring the timeliness and accuracy of the sorting action. This forms a complete automated process from detection to classification, significantly improving overall processing efficiency and ensuring the orderly sorting of products. In summary, the technical solution of this application achieves efficient and accurate automated detection of ultra-narrow-sized Kovar rings and improves the level of product quality control.

[0045] The various embodiments of this application have been described above. These descriptions are exemplary and not exhaustive, nor are they limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is chosen to best explain the principles, practical application, or improvement of the technology in the market, or to enable others skilled in the art to understand the embodiments disclosed herein.

Claims

1. A method of Kovar ring automatic detection sorting, characterized in that, The method comprises: The Kovar ring is fed to the work station turntable (2) through the vibration disc (3) and the vibration feeding slide rail (4), and the de-stacking operation and the positive and negative detection operation are performed during the feeding process, so as to ensure that only a single Kovar ring with the upper surface upward enters the work station turntable (2); The Kovar ring is transported by the work station turntable (2), and the Kovar ring is positioned in parallel with the tangential direction of the work station turntable (2) by using the guide plate; During the transportation of the Kovar ring, the images of the Kovar ring in multiple directions are collected by multiple vision cameras; The controller (12) receives the images and performs pre-processing, feature extraction and defect judgment to generate a detection result; Based on the detection result, the Kovar ring is sorted into qualified products, unqualified products and products to be rechecked by the sorting mechanism (11).

2. The method of claim 1, wherein the step of automatically detecting and sorting the vanadium ring is characterized by, The de-stacking operation and the positive and negative detection operation during the feeding process comprise: a plurality of de-stacking blocks arranged on the spiral track in the vibration disc (3) are used to make the overlapped Kovar rings slide back to the bottom of the vibration disc (3) under the action of gravity through the sliding guide block; The reflection type optical fiber probe is used to detect the surface reflectivity of the Kovar ring, and when the back surface of the Kovar ring is upward, the air jet controller (12) controls the air jet guide pipe to jet out gas to blow the Kovar ring back to the bottom of the vibration disc (3).

3. The automatic detection and sorting method for Kovar rings as described in claim 2, characterized in that, After the incoming Kovar ring is detected by the incoming sensor (5), the controller (12) calculates the time when the Kovar ring reaches each vision camera based on the rotation speed and distance of the work station turntable (2), and controls the vision camera to delay shooting.

4. The method of claim 1, wherein the step of automatically detecting and sorting the vanadium ring is characterized by, When the sorting mechanism (11) sorts the Kovar ring into qualified products, unqualified products and products to be rechecked, the controller (12) controls the air gun to delay starting for sorting based on the rotation speed of the work station turntable (2) and the distance between the fourth vision camera (10) and the sorting mechanism (11).

5. The automatic detection and sorting method for Kovar rings as described in claim 1, characterized in that, The defect judgment comprises: the gray scale segmentation and morphological processing of the pre-processed image are performed by using the VisionPro spot tool, the area, centroid and edge integrity feature parameters of the Kovar ring are extracted, and the feature parameters are compared with the standard parameter library to complete the quality judgment.

6. The automatic detection and sorting method for Kovar rings as described in claim 1, characterized in that, When the controller (12) pre-processes the received images, a gradient weighted median filtering algorithm adaptive to the ultra-narrow frame of the Kovar ring is adopted, which comprises: The gradient value of each pixel point in the image is calculated; The window size and weight of the median filtering are dynamically adjusted according to the gradient value of each pixel point in the image.

7. The method of claim 1 or 6, wherein the step of automatically detecting and sorting the vanadium ring is characterized by, In the feature extraction, the controller (12) performs multi-scale contour fusion analysis, which comprises: Different scale edge detection is performed on the binary Kovar ring image to extract multi-level contour features from coarse to fine; The multi-level contour features are fused based on the continuity of the contour curvature to reconstruct the Kovar ring edge contour with high integrity.

8. A Kovar ring auto-detection sorting apparatus, characterized by, It comprises: A machine table (1); A work station turntable (2) arranged at the center of the machine table (1) and transmissionally connected with a rotating motor in the machine table (1) for transporting the Kovar ring; A vibration disc (3) arranged on the machine table (1) for storing and feeding the Kovar ring; A vibration feeding slide rail (4) connected with the discharge port of the vibration disc (3) and the work station turntable (2) for guiding the Kovar ring; A come-in inductor (5) is arranged above the work station turntable (2) to detect the incoming Kovar ring; A plurality of visual cameras are fixedly arranged along the rotation direction of the work station turntable (2) to collect images of the Kovar ring; A sorting mechanism (11) is used to sort the Kovar ring based on the detection result; A controller (12) is arranged inside the machine table (1) and is electrically connected with the come-in inductor (5), the plurality of visual cameras and the sorting mechanism (11) to control the detection and sorting process.

9. The apparatus of claim 8, wherein, The inner wall of the vibrating disc (3) is provided with a spiral track rising spirally, a plurality of de-overlapping blocks are mounted on the track, the de-overlapping blocks are protruded relative to the inner wall of the vibrating disc (3) and the bottom is provided with a sliding guide block.

10. The apparatus of claim 9, wherein, A Kovar ring positive and negative detection assembly is arranged above the de-overlapping block, which comprises a reflective optical fiber probe, a jet guide pipe, a control valve and a jet controller (12), the jet controller (12) is electrically connected with the reflective optical fiber probe and the control valve.

Images