A stage contamination self-checking method and system

Abstract

The application discloses a kind of stage contamination self-checking method and system, method includes: based on the preset period acquisition machine empty state under the stage reference image, the validity of the stage reference image is judged;With pre-stored standard map, the contamination feature of the stage reference image is extracted by comparing effective stage reference image;The contamination feature is graded and determined, and the contamination feature after grading is adaptively compensated, and the contamination data is counted and traced to archive.The application effectively reduces the workload of re-trial, reduces the risk of product scrapping, improves the long-term detection stability of AOI equipment and the overall productivity of workshop.

Description

Technical Field

[0001] This invention relates to the field of LED chip manufacturing, and more specifically to a method and system for self-inspection of contamination on a substrate. Background Technology

[0002] As semiconductor manufacturing processes continue to iterate towards refinement, miniaturization, and high yield, the requirements for the accuracy and stability of chip appearance defect detection are constantly increasing. Automated Optical Inspection (AOI) technology, with its advantages of non-contact, high precision, and automation, has become a core technology for defect screening in semiconductor wafer and chip packaging processes. The stage, as the core carrier and benchmark component of semiconductor AOI equipment, is crucial for ensuring optical imaging quality and accurate defect detection due to its surface cleanliness, flatness, and optical consistency. If the stage has issues such as dirt or abnormal lighting, it will directly cause imaging deviations, leading to misjudgments. Therefore, precise control of the stage's condition is a core management point in the current field of high-precision semiconductor AOI inspection.

[0003] Currently, mainstream semiconductor AOI inspection equipment generally adopts a core inspection architecture that uses a stage to support the chip source and a dedicated light source and high-definition camera for imaging. By acquiring optical images of the chip surface and comparing them with standard image templates, it can automatically identify and detect various appearance defects such as chip scratches, stains, and damage.

[0004] However, existing mass-produced AOI equipment lacks intelligent self-inspection and adaptive control functions for its worktables. Worktable cleanliness control relies entirely on a passive maintenance model of manual periodic wiping and point-to-point inspections. Equipment detection parameters are fixed preset parameters and cannot be dynamically adjusted based on the real-time status of the worktables. Furthermore, existing equipment lacks modules for worktable contamination identification, grading, and data traceability. In actual mass production operations, only manual detection of obvious contamination allows for unified cleaning after shutdown, lacking differentiated treatment and intelligent compensation mechanisms—a common technical solution for current semiconductor mass production AOI equipment. Summary of the Invention

[0005] Based on this, the purpose of the present invention is to provide a method and system for self-inspection of stage contamination, which aims to solve the problems of difficulty in detecting contamination on AOI stages, easy misdiagnosis, and inflexible maintenance.

[0006] To achieve the above objectives, this invention proposes a self-inspection method for stage contamination, the method comprising: The validity of the platform reference image is determined by acquiring the platform reference image under no-load conditions at a preset period. The effective reference image of the platform is compared with the pre-stored standard image to extract the dirt features of the reference image of the platform. The dirt features are classified and determined, and adaptive compensation is performed on the classified dirt features. The dirt data is statistically analyzed and archived for source tracing.

[0007] According to one aspect of the above technical solution, the step of acquiring a platform reference image under no-load conditions based on a preset period and determining the validity of the platform reference image includes: Read the source image model code, capture reference images of the platform under no-load conditions based on a preset cycle, and archive them by model name; The reference image of the platform is converted into a grayscale image, the grayscale value of the reference image is extracted, and compared with the historical reference image to obtain the grayscale mean difference. If the grayscale mean difference reaches the preset reference value, it is confirmed that the platform is not heavily contaminated, and the reference image of the platform is retained according to the current machine parameters.

[0008] According to one aspect of the above technical solution, the step of comparing the effective stage reference image with a pre-stored standard image to extract the dirt features of the stage reference image includes: Gaussian filtering is used to eliminate minor noise in the grayscale image. The reference image of the platform is compared point by point with the pre-stored standard image to identify pixels with significant brightness differences. A preset brightness difference threshold is set. If the difference exceeds the brightness difference threshold, the pixels that exceed the brightness difference threshold are marked as suspected dirt. Connect all the pixels suspected of being dirty to obtain the complete suspected dirty area. Extract the pollution index of the suspected dirty area, remove small random noise, and obtain the real dirty area. Simultaneously obtain the dirt coordinates, shape and size, and feature parameters of the real dirty area. The feature parameters include dirt area, average gray level difference, and edge gradient.

[0009] According to one aspect of the above technical solution, the step of connecting all the suspected dirty pixels to obtain the complete suspected dirty area includes: The grayscale difference of each pixel is calculated between the acquired platform reference image and the pre-stored standard image. Based on the preset reference value, all suspected dirty discrete pixels that exceed the threshold are filtered out and a binary pixel mask layer is generated. Based on the minimum imaging size of the dirty stage, a rectangular structural element is selected. The rectangular structural element is used to perform an expansion operation on the pixel mask layer, so that adjacent scattered suspected pixels expand outward, connect and merge with each other, and connect the scattered independent pixels to form the outline of the dirty area. The same-sized structuring element erosion operation is performed on the outline of the dirty area to restore the true boundary range of the dirty area, and tiny connected regions with an area smaller than a set pixel threshold are removed to complete the division of the dirty block.

[0010] According to one aspect of the above technical solution, the steps of classifying and determining the dirt characteristics, adaptively compensating for the classified dirt characteristics, statistically analyzing the dirt data, and tracing and archiving it include: The actual dirty areas are divided according to their characteristic parameters to obtain the degree of dirtiness of the actual dirty areas. The degree of dirtiness includes slight dirtiness that does not affect production and heavy dirtiness. For areas with slight dirt, adaptive compensation is performed. Based on the dirt coordinates of the areas with slight dirt, a dirt marking layer is divided to distinguish between the dirty area, the clean stage background, and the effective area for chip detection. The marked dirty areas are then processed to achieve uniform brightness and eliminate stains. Extract the outline edge of the dirty area. If the dirty area overlaps with the chip edge, reduce the correction intensity of that area to fully preserve the true defect outline of the effective chip detection area.

[0011] According to one aspect of the above technical solution, the steps following the acquisition of the true defect contour of the effective detection area of ​​the chip are as follows: Perform overall brightness correction on the base image of the platform to eliminate local background color gradients caused by long-term use of the platform and unify the brightness standard of the entire image. After the correction is completed, compare it with the pre-saved standard image again. If the brightness difference caused by the stain reaches the preset percentage standard, the compensation is deemed effective and a record is saved. If the compensation effect is insufficient, it is automatically upgraded to be heavily dirty and a cleaning reminder is pushed. Collect and archive data on platform dirt and calibration.

[0012] According to one aspect of the above technical solution, the step of extracting the contour edge of the dirty area includes: A pre-stored standard chip contour template is used to obtain a stage reference image, and then the effective chip detection area in the stage reference image is locked. Using edge detection operators, we filter out interfering pixels with internal chip textures, extract the outer contour lines of the chip, generate a chip edge contour layer, and mark and store it. The dirt mark layer and the chip edge contour layer are overlaid and compared to determine whether the dirt area overlaps with the chip contour area. If the two areas overlap, the image correction intensity of the overlapping area is dynamically reduced according to the overlap area ratio.

[0013] This invention also proposes a stage contamination self-inspection system, which is used to implement the above-mentioned stage contamination self-inspection method. The system includes: The determination module is used to acquire a reference image of the platform under no-load conditions based on a preset period, and to determine the validity of the reference image of the platform. The extraction module is used to compare the valid platform reference image with the pre-stored standard image and extract the dirt features of the platform reference image; The compensation module is used to classify and determine the dirt features, adaptively compensate the classified dirt features, statistically analyze the dirt data, and archive the data for source tracing.

[0014] The present invention also proposes a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described platform contamination self-check method.

[0015] The present invention also proposes an electronic device, including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the computer program to implement the above-described platform dirt self-check method.

[0016] In summary, the stage contamination self-inspection method provided by this invention can automatically collect and periodically update the stage reference image, avoiding reference inaccuracies caused by light source attenuation and stage oxidation. It relies on multi-dimensional identification of various types of contamination, such as dust and adhesive residue, based on grayscale difference, area, and reflectivity gradient, and performs light and heavy classification. Furthermore, for minor contamination, a layered image adaptive compensation algorithm is activated to distinguish between stage contamination interference and actual chip defects, automatically eliminating false defect features. Cleaning reminders are only pushed to heavily contaminated areas during equipment idle intervals, significantly reducing unnecessary downtime and false positives for good products. Simultaneously, it comprehensively archives all dimensions of contamination data, supports multi-condition traceability, automatically generates maintenance reports, and integrates with the MES system. It can accurately summarize stage contamination patterns to standardize cleaning and maintenance cycles, balancing detection accuracy, production line efficiency, and standardized equipment maintenance. It can also independently configure judgment thresholds for different chip models, quickly adapting to multi-model switching mass production scenarios. This invention effectively reduces the workload of re-judgment, lowers the risk of product scrap, and improves the long-term detection stability of AOI equipment and the overall production capacity of the workshop.

[0017] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0018] Figure 1 This is a flowchart of the platform contamination self-inspection method in Embodiment 1 of the present invention; Figure 2 This is a flowchart of the platform contamination self-inspection method in Embodiment 2 of the present invention. Figure 3 This is a flowchart illustrating the AOI stage contamination self-inspection and image compensation execution process in Embodiment 2 of the present invention. Figure 4 This is a schematic diagram of the two-level classification logic for platform contamination in Embodiment 2 of the present invention; Figure 5 This is a schematic diagram of the platform contamination self-inspection system in Embodiment 3 of the present invention; Figure 6 This is a structural block diagram of the electronic device in Embodiment 5 of the present invention. Detailed Implementation

[0019] To facilitate understanding of the present invention, a more complete description will be given below with reference to the accompanying drawings. Several embodiments of the invention are illustrated in the drawings. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete.

[0020] It should be noted that when a component is said to be "fixed to" another component, it can be directly on the other component or there may be an intervening component. When a component is said to be "connected to" another component, it can be directly connected to the other component or there may be an intervening component. The terms "vertical," "horizontal," "left," "right," and similar expressions used in this document are for illustrative purposes only.

[0021] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0022] Example 1 like Figure 1 The diagram shows a flowchart of a stage contamination self-inspection method according to Embodiment 1 of the present invention. The stage contamination self-inspection method includes the following steps S11-S13, wherein: S11. Obtain a reference image of the platform under no-load conditions based on a preset period, and determine the validity of the reference image of the platform.

[0023] Every day when the equipment is turned on or when changing production models, it automatically takes a picture of a clean platform under no-load conditions, which serves as the platform's reference image. After prolonged use, the platform will experience normal slight oxidation and background color changes. The system will also automatically update the platform's reference image periodically to avoid inaccurate judgments due to outdated reference images. All reference data is automatically saved, providing a standard basis for subsequent comparisons of dirt and grime.

[0024] Each time a new platform reference image is acquired, its validity is determined based on the difference in the average grayscale value of the platform reference image. If valid, it is automatically saved; if invalid, the image is retaken until a platform reference image acquired in real time is deemed valid.

[0025] S12. Compare the valid platform reference image with the pre-stored standard image to extract the dirt features of the platform reference image.

[0026] During the intervals between batches of testing or when the system is idle, without placing any LED chip sources on the stage, the system automatically turns on the light source to illuminate the stage and captures a real-time image of the empty stage. This real-time image of the stage is then compared with a standard clean image. The system automatically identifies dust, solder dross, stains, and reflective patches on the stage by analyzing grayscale differences, pixel variations, and edge anomalies, while simultaneously recording the location, area, and degree of interference of the contaminants.

[0027] S13. The dirt features are classified and determined, and adaptive compensation is performed on the classified dirt features. The dirt data is statistically analyzed and archived for source tracing.

[0028] The system sets up a two-level dirt assessment standard, eliminating the need to shut down the machine for every instance of dirt. Minor dirt, with its small area and low grayscale variation, does not affect the identification of actual defects, and the system automatically initiates the image compensation process. Severe dirt, with its large area and significant reflections and shadows, cannot be completely eliminated by algorithm correction; the system will then display a pop-up reminder to the technician to clean the equipment during idle periods, ensuring uninterrupted production.

[0029] For minor contamination that will not affect production, the equipment will automatically activate the image correction function. By balancing image grayscale, eliminating local noise, and correcting uneven lighting, it filters out false defect interference caused by stage contamination. During the correction process, the equipment retains the edges and features of the chip's actual defects, eliminating only the interference caused by the stage, ensuring accurate detection without requiring machine downtime.

[0030] Each time the equipment detects dirt on the platform, it automatically records data, including the detection time, platform number, dirt level, whether compensation was required, and whether a cleaning reminder was issued. It automatically generates a simple maintenance report and saves it to the backend. Administrators can view the frequency of dirt accumulation on each platform at any time, summarize cleaning patterns, and make daily maintenance more scientific and standardized.

[0031] In summary, the stage contamination self-inspection method provided by this invention can automatically collect and periodically update the stage reference image, avoiding reference inaccuracies caused by light source attenuation and stage oxidation. It relies on multi-dimensional identification of various types of contamination, such as dust and adhesive residue, based on grayscale difference, area, and reflectivity gradient, and performs light and heavy classification. Furthermore, for minor contamination, a layered image adaptive compensation algorithm is activated to distinguish between stage contamination interference and actual chip defects, automatically eliminating false defect features. Cleaning reminders are only pushed to heavily contaminated areas during equipment idle intervals, significantly reducing unnecessary downtime and false positives for good products. Simultaneously, it comprehensively archives all dimensions of contamination data, supports multi-condition traceability, automatically generates maintenance reports, and integrates with the MES system. It can accurately summarize stage contamination patterns to standardize cleaning and maintenance cycles, balancing detection accuracy, production line efficiency, and standardized equipment maintenance. It can also independently configure judgment thresholds for different chip models, quickly adapting to multi-model switching mass production scenarios. This invention effectively reduces the workload of re-judgment, lowers the risk of product scrap, and improves the long-term detection stability of AOI equipment and the overall production capacity of the workshop.

[0032] Example 2 Please see Figure 2 The figure shows a self-inspection method for stage contamination in the second embodiment of the present invention, including steps S21-S23.

[0033] S21. Obtain a reference image of the platform under no-load conditions based on a preset period, and determine the validity of the reference image of the platform.

[0034] When the equipment is turned on each day or the production model is changed, the source film model code is read, and the reference image of the platform is taken at no load based on the preset period as the standard reference image, and archived by model name.

[0035] After prolonged use, the platform will experience normal slight oxidation and background color changes. The system will also automatically update the baseline image periodically to avoid inaccurate judgments due to outdated reference images. All baseline data is automatically saved to provide a standard basis for subsequent dirt comparison.

[0036] Specifically, in this embodiment, the device is powered on and initialized, the stage performs a full-width homing to the no-load safe position, the ring coaxial light source switches to standard brightness (LED power 85%, exposure value 120μs), and the 2K industrial CMOS camera captures a complete image of the stage with a resolution of 2560×1920.

[0037] Switching LED chip models: The system reads the chip source model code, automatically clears the old model reference library, takes a new reference image under no-load conditions, and archives it by model name. It automatically updates at set intervals. By setting the preset cycle to an optional cycle of 2h / 4h / 8h, when the cycle is reached and the equipment is in no-load standby (no wafer loading for 30 consecutive seconds), the reference image is automatically retaken.

[0038] The reference image of the platform is converted into a grayscale image, and a core quantization metric ΔG (grayscale mean difference) is introduced: Δ represents the difference, and G represents the image grayscale. ΔG is used to measure the overall brightness difference between the newly captured empty platform image and the historical reference image. The grayscale values ​​of the reference image are extracted and compared with the historical reference image to obtain the grayscale mean difference. ΔGG<5: The background color of the platform does not change significantly, and the original reference is retained; 5≤ΔG≤15: Slight stage oxidation / light source attenuation, automatically fuse old and new images to generate a new benchmark. ΔG > 15: The reference is invalid. A pop-up window prompts you to manually confirm that the stage is not heavily dirty and save a new reference.

[0039] The local hard drive stores data in folders based on machine number, date, and model, retaining the baseline images for the most recent 30 days. Images exceeding this limit are automatically overwritten, and manual export backup is supported.

[0040] S22. Compare the valid platform reference image with the pre-stored standard image to extract the dirt features of the platform reference image.

[0041] After converting the platform reference image into a grayscale image, a 3×3 Gaussian filter can be used to eliminate the small noise generated by the camera sensor. The system compares the stage reference image with the pre-stored standard image point by point to identify pixels with significant brightness differences. A brightness difference threshold is set. If the brightness difference between the pixels in the stage reference image and the pre-stored standard image exceeds the brightness difference threshold, the pixels in the stage reference image are marked as suspected dirt.

[0042] By employing a 3×3 dilation operation, scattered suspected pixels are connected into a complete dirty area. Three key indicators for each dirty area are statistically analyzed: dirty area S (number of pixels), average gray level difference ΔG_avg, and edge gradient Grad. ΔG_avg represents the difference between the overall brightness of a single dirty area and the reference image, which is used to determine the strength of the interference between light and dark in the dirty area; Grad represents the magnitude of the change in light and dark at the edge of the area, which is used to identify highly reflective dirt.

[0043] It should be noted that in the step of connecting all the suspected dirty pixels to obtain the complete suspected dirty area, the gray-level difference of the acquired platform reference image and the pre-stored standard image is calculated pixel by pixel. Based on the set preset reference value, all suspected dirty discrete pixels that exceed the threshold are filtered out to generate a binary pixel mask layer. Based on the minimum imaging size of the dirt on the platform, a rectangular structural element is selected, with a radius typically set to 3-5 pixels. This rectangular structural element is used to perform a dilation operation on the pixel mask layer, causing adjacent scattered suspected pixels to expand outwards, connect and merge, thus connecting the scattered, independent pixels to form a continuous, closed outline of the dirt region. The same-size structuring element erosion operation is performed on the outline of the dirty area to offset the problem of area outline magnification caused by expansion, restore the true boundary range of the dirty area, remove the redundant noise area caused by excessive expansion, and at the same time remove the tiny connected regions with an area smaller than the set pixel threshold, filter the isolated noise generated by the camera sensor, retain the continuous dirty connected regions, and complete the dirty block division.

[0044] After connecting all the suspected dirty pixels, the complete suspected dirty area is obtained. The pollution index of the suspected dirty area is extracted, and small random noise is removed to obtain the real dirty area. The dirt coordinates, shape and size and feature parameters of the real dirty area are obtained at the same time. The feature parameters include dirt area, average gray level difference and edge gradient.

[0045] After traversing the complete dirty region after connectivity, the total number of all valid pixels in the region is counted. Combined with the physical size of the camera pixels, the actual physical area of ​​the dirty region is calculated. The gray values ​​of all pixels in the dirty region are extracted, and the average pixel value of the dirty region and the average pixel value of the corresponding reference empty image are calculated. The difference between the two is calculated to obtain the average gray value difference of the dirty region. The Sobel operator is used to extract the gray value gradient value of the contour edge of the dirty region, and the average gradient value of all positions of the edge is calculated. The higher the Grad value, the stronger the contrast between light and dark at the dirty edge and the more obvious the dirty features.

[0046] S23. The dirt features are classified and determined, and adaptive compensation is performed on the classified dirt features. The dirt data is statistically analyzed and archived for source tracing.

[0047] S231. Grading of dirt and grime severity.

[0048] The degree of dirtiness of the actual dirty area is obtained by dividing the area according to the characteristic parameters of the actual dirty area. The degree of dirtiness includes slight dirtiness that does not affect production and heavy dirtiness.

[0049] For minor contamination to be determined, two conditions must be met simultaneously: the area of ​​a single contaminated connected region S ≤ 120 pixels, and the average grayscale difference ΔG_avg ≤ 22, with a reflectivity gradient Grad < 35. Since minor contamination has a low level of interference and will not cover the chip's detection area, the image compensation process will automatically proceed.

[0050] Severe soiling is determined by meeting any one of the following criteria: The area of ​​a single dirty patch, S, is greater than 120 pixels. ΔG_avg>22, strong reflectivity gradient Grad≥35; Three or more minor stains are superimposed on the same image; Since adaptive compensation cannot effectively address heavily soiled areas, it can only provide pop-up reminders and push cleaning reminders during the idle intervals of the equipment, without interrupting the testing of the current batch.

[0051] It should be noted that a buffer threshold is set in the range of pixel area S=100~120 pixels. Only when two consecutive frames are identified as dirty will a classification be determined, in order to avoid misclassification due to fluctuations in a single imaging.

[0052] S232, Adaptive compensation for dirty images of the platform.

[0053] For areas with slight dirt, adaptive compensation is performed. Based on the dirt coordinates of the areas with slight dirt, a dirt marking layer is divided to distinguish between the dirty area, the clean stage background, and the effective area for chip detection. The marked dirty areas are then processed to uniformly increase brightness and eliminate bright and dark spots caused by the dirt.

[0054] Extract the outline edge of the contaminated area. If the contaminated area overlaps with the chip edge, reduce the correction intensity of that area to fully preserve the true defect outline of the effective chip detection area. Specifically: A standard chip outline template, including chip dimensions and edge coordinate data, is pre-entered; after the device acquires real-time images, a template matching algorithm is used to locate the precise position of the chip within the current frame, locking the entire chip area. An edge extraction algorithm is applied to the chip area. The Canny edge detection operator is used, and high and low thresholds are set to filter out interference pixels of internal chip texture. The outer contour of the chip and the internal chip edge and scratch contour lines are accurately captured, and a chip edge contour layer is generated and marked and stored. Compare the dirt marking layer with the chip edge outline layer to determine whether the dirt area overlaps with the chip outline area. If the two areas overlap, the image correction intensity of the overlapping area will be automatically reduced. The correction intensity of the overlapping area will be dynamically adjusted according to the ratio of the overlapping area to preserve the true defect features of the chip edge to the greatest extent. Only the pure stage dirty area will be subjected to full brightness equalization processing.

[0055] After obtaining the true defect outline of the effective area for chip detection, the overall brightness of the entire image is corrected to eliminate local background color gradients caused by long-term use of the platform and to unify the brightness standard of the entire image. After the correction is completed, the image is compared with the reference image again. If the brightness difference caused by the stain is reduced by more than 70%, the compensation is deemed effective and recorded. If the compensation effect is insufficient, the image is automatically upgraded to be heavily soiled and a cleaning reminder is sent.

[0056] S233, Sewage Data Statistics and Source Tracing Archiving.

[0057] Each time the equipment detects dirt on the platform, it automatically records data, including the detection time, platform number, dirt level, whether compensation was required, and whether a cleaning reminder was issued. It automatically generates a simple maintenance report and saves it to the backend. Administrators can view the frequency of dirt accumulation on each platform at any time, summarize cleaning patterns, and make daily maintenance more scientific and standardized.

[0058] The complete storage fields for a single dirt log include machine ID, production model, timestamp, dirt coordinates, area S, grayscale difference ΔG_avg, classification result, whether compensation is required, compensation effect, whether pop-up reminder is required, and processing time. The dimensions automatically calculated in the operation and maintenance report include daily / weekly / monthly dirt occurrence frequency, average dirt accumulation time for each product model, percentage of heavily dirty products, compensation effectiveness rate, and average number of production units between two cleaning cycles.

[0059] The stored data supports multi-condition retrieval by machine, time, and product model, and can be exported as a CSV log with one click. It can be integrated with the MES production line system to automatically push cleaning early warning work orders.

[0060] In summary, the stage contamination self-inspection method provided by this invention can automatically collect and periodically update the stage reference image, avoiding reference inaccuracies caused by light source attenuation and stage oxidation. It relies on multi-dimensional identification of various types of contamination, such as dust and adhesive residue, based on grayscale difference, area, and reflectivity gradient, and performs light and heavy classification. Furthermore, for minor contamination, a layered image adaptive compensation algorithm is activated to distinguish between stage contamination interference and actual chip defects, automatically eliminating false defect features. Cleaning reminders are only pushed to heavily contaminated areas during equipment idle intervals, significantly reducing unnecessary downtime and false positives for good products. Simultaneously, it comprehensively archives all dimensions of contamination data, supports multi-condition traceability, automatically generates maintenance reports, and integrates with the MES system. It can accurately summarize stage contamination patterns to standardize cleaning and maintenance cycles, balancing detection accuracy, production line efficiency, and standardized equipment maintenance. It can also independently configure judgment thresholds for different chip models, quickly adapting to multi-model switching mass production scenarios. This invention effectively reduces the workload of re-judgment, lowers the risk of product scrap, and improves the long-term detection stability of AOI equipment and the overall production capacity of the workshop.

[0061] Example 3 In another aspect, the present invention also provides a platform contamination self-inspection system, please refer to [link / reference needed]. Figure 5 The diagram shown is a structural schematic of the platform contamination self-inspection system in Embodiment 2 of the present invention. The platform contamination self-inspection system includes: The determination module 11 is used to acquire a reference image of the platform under no-load conditions based on a preset period and determine the validity of the reference image of the platform. Extraction module 12 is used to compare the valid stage reference image with the pre-stored standard image and extract the dirt features of the stage reference image; The compensation module 13 is used to classify and determine the dirt features, perform adaptive compensation on the classified dirt features, collect dirt data, and archive the data for tracing.

[0062] Every day when the equipment is turned on or when changing production models, it automatically takes a picture of a clean platform under no-load conditions, which serves as the platform's reference image. After prolonged use, the platform will experience normal slight oxidation and background color changes. The system will also automatically update the platform's reference image periodically to avoid inaccurate judgments due to outdated reference images. All reference data is automatically saved, providing a standard basis for subsequent comparisons of dirt and grime.

[0063] Each time a new platform reference image is acquired, its validity is determined based on the difference in the average grayscale value of the platform reference image. If valid, it is automatically saved; if invalid, the image is retaken until a platform reference image acquired in real time is deemed valid.

[0064] During the intervals between batches of testing or when the system is idle, without placing any LED chip sources on the stage, the system automatically turns on the light source to illuminate the stage and captures a real-time image of the empty stage. This real-time image of the stage is then compared with a standard clean image. The system automatically identifies dust, solder dross, stains, and reflective patches on the stage by analyzing grayscale differences, pixel variations, and edge anomalies, while simultaneously recording the location, area, and degree of interference of the contaminants.

[0065] The system sets up a two-level dirt assessment standard, eliminating the need to shut down the machine for every instance of dirt. Minor dirt, with its small area and low grayscale variation, does not affect the identification of actual defects, and the system automatically initiates the image compensation process. Severe dirt, with its large area and significant reflections and shadows, cannot be completely eliminated by algorithm correction; the system will then display a pop-up reminder to the technician to clean the equipment during idle periods, ensuring uninterrupted production.

[0066] For minor contamination that will not affect production, the equipment will automatically activate the image correction function. By balancing image grayscale, eliminating local noise, and correcting uneven lighting, it filters out false defect interference caused by stage contamination. During the correction process, the equipment retains the edges and features of the chip's actual defects, eliminating only the interference caused by the stage, ensuring accurate detection without requiring machine downtime.

[0067] Each time the equipment detects dirt on the platform, it automatically records data, including the detection time, platform number, dirt level, whether compensation was required, and whether a cleaning reminder was issued. It automatically generates a simple maintenance report and saves it to the backend. Administrators can view the frequency of dirt accumulation on each platform at any time, summarize cleaning patterns, and make daily maintenance more scientific and standardized.

[0068] In summary, the platform contamination self-inspection system provided by this invention can automatically collect and periodically update the platform reference image, avoiding reference inaccuracies caused by light source attenuation and platform oxidation. It identifies various types of contamination, such as dust and adhesive residue, based on grayscale difference, area, and reflectivity gradient, and classifies them by severity. Furthermore, for minor contamination, it activates a layered image adaptive compensation algorithm to distinguish between platform contamination interference and actual chip defects, automatically eliminating false defect features. Cleaning reminders are only pushed to heavily contaminated areas during equipment idle intervals, significantly reducing unnecessary downtime and false positives for good products. Simultaneously, it comprehensively archives all dimensions of contamination data, supports multi-condition traceability, automatically generates maintenance reports, and integrates with the MES system. It can accurately summarize platform contamination patterns to standardize cleaning and maintenance cycles, balancing detection accuracy, production line efficiency, and standardized equipment maintenance. It can also independently configure judgment thresholds for different chip models, quickly adapting to multi-model switching mass production scenarios. This invention effectively reduces the workload of re-judgment, lowers the risk of product scrap, and improves the long-term detection stability of AOI equipment and the overall production capacity of the workshop.

[0069] Example 4 In another aspect, the present invention also proposes a computer-readable storage medium having stored thereon one or more computer programs that, when executed by a processor, implement the aforementioned platform contamination self-checking method.

[0070] Those skilled in the art will understand that the logic or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable storage medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable storage medium" can mean any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0071] More specific examples (a non-exhaustive list) of computer-readable storage media include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable storage media can even be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0072] Example 5 Figure 6 This is a structural block diagram of an electronic device provided in Embodiment 4. The electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the platform contamination self-checking method in the above embodiments. Figure 6 The electronic device 30 shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments of the present invention.

[0073] like Figure 6 As shown, the electronic device 30 can be manifested as a general-purpose computing device, such as a server device. The components of the electronic device 30 may include, but are not limited to: at least one processor 31, at least one memory 32, and a bus 33 connecting different system components (including memory 32 and processor 31).

[0074] Bus 33 includes a data bus, an address bus, and a control bus.

[0075] The memory 32 may include volatile memory, such as RAM 321 (random access memory), and / or cache memory 322, and may further include ROM 323 (read-only memory).

[0076] The memory 32 may also include a program tool 325 having a set (at least one) of program modules 324, including but not limited to: an operating system, one or more application programs, other program modules, and program data, each or some combination of these examples may include an implementation of a network environment.

[0077] The processor 31 executes various functional applications and data processing by running computer programs stored in the memory 32, such as the platform dirt self-checking method of the present invention as described above.

[0078] Electronic device 30 can also communicate with one or more external devices 34 (e.g., keyboard, pointing device, etc.). This communication can be performed via I / O interface 35 (input / output interface). Furthermore, electronic device 30 can also communicate with one or more networks (e.g., local area network (LAN), wide area network (WAN), and / or public network, such as the Internet) via network adapter 36. Figure 6 As shown, network adapter 36 communicates with other modules of the model-generated electronic device 30 via bus 33. It should be understood that, although not shown in the figure, other hardware and / or software modules can be used in conjunction with the model-generated electronic device 30, including but not limited to: microcode, device drivers, redundant processors, disk drive arrays, RAID (disk array) systems, tape drives, and data backup storage systems.

[0079] It should be noted that although several units / modules or sub-units / modules of the electronic device have been mentioned in the detailed description above, this division is merely exemplary and not mandatory. In fact, according to embodiments of the present invention, the features and functions of two or more units / modules described above can be embodied in one unit / module. Conversely, the features and functions of one unit / module described above can be further divided and embodied by multiple units / modules.

[0080] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0081] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

Claims

1. A method for self-inspection of platform contamination, characterized in that, The methods include: The validity of the platform reference image is determined by acquiring the platform reference image under no-load conditions at a preset period. The effective reference image of the platform is compared with the pre-stored standard image to extract the dirt features of the reference image of the platform. The dirt features are classified and determined, and adaptive compensation is performed on the classified dirt features. The dirt data is statistically analyzed and archived for source tracing.

2. The self-inspection method for stage contamination according to claim 1, characterized in that, The step of acquiring a reference image of the platform under no-load conditions based on a preset period and determining the validity of the reference image includes: Read the source image model code, capture reference images of the platform under no-load conditions based on a preset cycle, and archive them by model name; The reference image of the platform is converted into a grayscale image, the grayscale value of the reference image is extracted, and compared with the historical reference image to obtain the grayscale mean difference. If the grayscale mean difference reaches the preset reference value, it is confirmed that the platform is not heavily contaminated, and the reference image of the platform is retained according to the current machine parameters.

3. The self-inspection method for stage contamination according to claim 1, characterized in that, The step of comparing the valid platform reference image with a pre-stored standard image to extract the dirt features of the platform reference image includes: Gaussian filtering is used to eliminate minor noise in the grayscale image. The reference image of the platform is compared point by point with the pre-stored standard image to identify pixels with significant brightness differences. A preset brightness difference threshold is set. If the difference exceeds the brightness difference threshold, the pixels that exceed the brightness difference threshold are marked as suspected dirt. Connect all the pixels suspected of being dirty to obtain the complete suspected dirty area. Extract the pollution index of the suspected dirty area, remove small random noise, and obtain the real dirty area. Simultaneously obtain the dirt coordinates, shape and size, and feature parameters of the real dirty area. The feature parameters include dirt area, average gray level difference, and edge gradient.

4. The self-inspection method for stage contamination according to claim 3, characterized in that, The step of connecting all the suspected dirty pixels to obtain the complete suspected dirty area includes: The grayscale difference of each pixel is calculated between the acquired platform reference image and the pre-stored standard image. Based on the preset reference value, all suspected dirty discrete pixels that exceed the threshold are filtered out and a binary pixel mask layer is generated. Based on the minimum imaging size of the dirty stage, a rectangular structural element is selected. The rectangular structural element is used to perform an expansion operation on the pixel mask layer, so that adjacent scattered suspected pixels expand outward, connect and merge with each other, and connect the scattered independent pixels to form the outline of the dirty area. The same-sized structuring element erosion operation is performed on the outline of the dirty area to restore the true boundary range of the dirty area, and tiny connected regions with an area smaller than a set pixel threshold are removed to complete the division of the dirty block.

5. The self-inspection method for stage contamination according to claim 1, characterized in that, The steps of classifying and determining the dirt features, adaptively compensating for the classified dirt features, statistically analyzing the dirt data, and tracing and archiving the data include: The actual dirty areas are divided according to their characteristic parameters to obtain the degree of dirtiness of the actual dirty areas. The degree of dirtiness includes slight dirtiness that does not affect production and heavy dirtiness. For areas with slight dirt, adaptive compensation is performed. Based on the dirt coordinates of the areas with slight dirt, a dirt marking layer is divided to distinguish between the dirty area, the clean stage background, and the effective area for chip detection. The marked dirty areas are then processed to achieve uniform brightness and eliminate stains. Extract the outline edge of the dirty area. If the dirty area overlaps with the chip edge, reduce the correction intensity of that area to fully preserve the true defect outline of the effective chip detection area.

6. The self-inspection method for stage contamination according to claim 5, characterized in that, The steps after obtaining the true defect contour of the effective detection area of ​​the chip are as follows: Perform overall brightness correction on the base image of the platform to eliminate local background color gradients caused by long-term use of the platform and unify the brightness standard of the entire image. After the correction is completed, compare it with the pre-saved standard image again. If the brightness difference caused by the stain reaches the preset percentage standard, the compensation is deemed effective and a record is saved. If the compensation effect is insufficient, it is automatically upgraded to be heavily dirty and a cleaning reminder is pushed. Collect and archive data on platform dirt and calibration.

7. The self-inspection method for stage contamination according to claim 5, characterized in that, The step of extracting the contour edge of the dirty area includes: A pre-stored standard chip contour template is used to obtain a stage reference image, and then the effective chip detection area in the stage reference image is locked. Using edge detection operators, we filter out interfering pixels with internal chip textures, extract the outer contour lines of the chip, generate a chip edge contour layer, and mark and store it. The dirt mark layer and the chip edge contour layer are overlaid and compared to determine whether the dirt area overlaps with the chip contour area. If the two areas overlap, the image correction intensity of the overlapping area is dynamically reduced according to the overlap area ratio.

8. A platform contamination self-inspection system, characterized in that, The stage contamination self-inspection system is used to implement the stage contamination self-inspection method according to any one of claims 1-7, the system comprising: The determination module is used to acquire a reference image of the platform under no-load conditions based on a preset period, and to determine the validity of the reference image of the platform. The extraction module is used to compare the valid platform reference image with the pre-stored standard image and extract the dirt features of the platform reference image; The compensation module is used to classify and determine the dirt features, adaptively compensate the classified dirt features, statistically analyze the dirt data, and archive the data for source tracing.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by the processor, the program implements the stage contamination self-check method as described in any one of claims 1-7.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the platform dirt self-check method as described in any one of claims 1-7.

Images