Connector multi-pin self-steady state detection method and system based on 3D machine vision

Abstract

The invention belongs to the technical field of machine vision, and discloses a connector multi-pin self-steady state detection method and system based on 3D machine vision, and the method comprises the following steps: automatically calculating the position information and height of each pin according to the data collected by a 3D camera, simulating the stable posture of a connector product on a plane through an algorithm, and carrying out the self-steady state detection. And constructing three most stable points of the plane support, attaching a weight value to each type of calculation parameters, calculating final scores of every three groups of pins, and obtaining three pins with the highest scores to establish a reference surface to calculate whether the product to be detected is qualified or not. According to the method, software and hardware collaboration is adopted, point cloud data processing, three-dimensional image modeling and other methods based on 3D images are integrated, connector type pins are processed and analyzed, the fitting performance, compatibility, stability and accuracy of the stability of each set of pins are higher in the calculation and analysis process, meanwhile, the problem that the yield is unstable in the production process is solved, and the production efficiency is improved. Detection efficiency is high, defect types are multiple, and application is wide.

Description

Technical Field

[0001] The present invention relates to the technical field of pin detection for connector elements, and in particular to a method and system for self-steady-state detection of multiple pins of a connector based on 3D machine vision. Background Art

[0002] With the development trend of today's society, the demand for electronic equipment continues to increase, which has led to the requirement for fast and efficient production of integrated circuits. In the process of rapid production of integrated circuits, it is very easy for the pins of components to be missing, broken, or deformed due to certain external factors. If the pins on the integrated circuit board have such defects, the product will be unusable or its service life will be affected. Therefore, in the production process of integrated circuits, it is particularly necessary to detect the flatness defects of the pins.

[0003] At present, enterprises mainly use manual visual inspection and traditional two-dimensional defect detection methods to detect crooked pins, broken pins, and deformed pins of circuit board components. This method not only has the problems of high labor intensity and low efficiency, but also the staff is very likely to make omissions during long-term high-intensity work, thus affecting the final result of pin defect detection. At present, there are also a few methods that use digital image processing to replace manual pin detection methods, but mainly in the process of non-contact optical detection of connector solder pin flatness, the placement of the product needs to simulate the state of being soldered to the circuit board during actual use. Single-row soldered connectors can use a jig board to simulate the circuit board, and the product is placed on the jig for detection, but multi-row pin connectors can only be detected by 3D scanning to avoid empty pins. At this time, there is no direct support reference surface, and only a fixed virtual plane can be used for detection, resulting in inaccurate measurement. There is no scientific algorithm to calculate and evaluate the stable posture of connector products in actual use, resulting in the inability to detect defective products and large measurement numerical errors. Summary of the invention

[0004] In order to solve the various defect problems existing in pin defect detection in the prior art, the present invention provides a connector multi-pin self-steady-state detection method and system based on 3D machine vision. Through the coordinated cooperation of software and hardware, especially after two center of gravity distance calculations, the stability of each group of pins is calculated and analyzed, and the fitting, compatibility, stability and accuracy are simultaneously improved. While improving the detection accuracy and efficiency, it also solves the problem of unstable yield in the production process.

[0005] The present invention provides the following technical solutions: A connector multi-pin self-steady-state detection method based on 3D machine vision includes the following steps: first, automatically calculating the position information and height of each pin according to the data collected by a 3D camera, simulating the stable support posture of the connector product on the plane through an algorithm, constructing the three most stable points of the plane support, and then attaching a weight value to each type of calculation parameter, calculating the final score of each three groups of pins, and establishing a reference plane to calculate the data by obtaining the three pins with the highest scores, and finally comparing with the product tolerance standard to determine whether the connector product to be tested is qualified, and judging that the product that exceeds the tolerance range is unqualified, and judging that the product that does not exceed the tolerance range is qualified.

[0006] The specific steps include: Step 1: Control the 3D scanning head to perform a full range of motion scanning on the surface of the first multi-pin connector product to be tested on the production line. During the scanning process, the laser line cross-sectional profile data is captured and recorded in real time; Step 2: The data collector converts the captured laser line cross-sectional profile data into height data according to the principle of laser triangulation, and the converted data forms a series of height profile data sets; Step 3: The height profile data set collected in step 2 is transmitted to the visual controller, which corrects and restores the received data to generate complete three-dimensional point cloud data; Step 4: Segment the features of the connector pins in the point cloud data according to the height range, and use the point cloud segmentation algorithm to detect the Z height and corresponding X, Y coordinates of each pin; Step 5: Traverse all the pin data, and every three pins form a combination; Step 6: Use the feature point detection algorithm to calculate the three reference points in each combination, and calculate the area of ​​the triangle formed by the three reference points according to formula 1 Area , and the fractional ratio S1 obtained by calculating the area term afterwards: Formula 1 Where: α is the weight of the area. After the area is normalized, it is directly used as a positive influence. The larger the area, the higher the score. Step 7: Based on the reference plane constructed in step 6, perform a gravity center distance calculation to calculate the center of the plane formed by each combination of pins and the Euclidean distance between the center of gravity of the product and this plane. Dis tan ce , and calculate the fractional ratio S2 of the centroid distance term according to formula 2: Formula 2 Where: β is the weight, which controls the influence of the distance from the center of gravity on the score; k1 is the amplification factor that adjusts the exponential growth rate; Step 8: Calculate the sum of the height differences of all the remaining pins outside each group projected onto the plane formed by the pins in this group. Height_Diff, And the fractional ratio S3 obtained by calculating the height difference term according to Formula 3: Formula 3 Where: γ is the weight, which controls the influence of height difference on the score; Step 9: Using the data calculated in steps 6-8, calculate the stability scores of any three pins that may be used to construct support points, and select the three pins with the highest stability scores as the support points, i.e., the reference points, for subsequent product simulations; Step 10: Use the calculated three reference points to construct the reference plane; Step 11: Recalculate the heights of all pins to the reference plane according to Formula 4, perform secondary gravity center distance calculation, and obtain flatness data S0; Formula 4 In the formula: 100 / α: means to ensure that the theoretical full score is 100, α is the weight of the area, corresponding to the score when the theoretical score is full; Step 12: Compare the calculated data of the first connector product to be tested with the product tolerance standard to determine whether the connector product is qualified. If the connector product exceeds the tolerance range, it is determined to be a defective product, and if the connector product does not exceed the tolerance range, it is determined to be a qualified product. Step 13: Repeat steps 1 to 12 to test each multi-pin connector product to be tested on the production line in turn until the production of this batch of multi-pin connector products is completed.

[0007] A connector multi-pin self-steady-state detection system based on 3D machine vision, which includes an image acquisition module, an image processing module, and a data calculation module, wherein the image acquisition module includes an industrial light source and an industrial camera, the image processing module includes a point cloud data processing program and a program for generating 3D pictures; the data calculation module includes an area calculation program, a center of gravity distance calculation program, a secondary center of gravity calculation program, and a score calculation program for each group of pins; The control program software part that implements the connector multi-pin self-steady-state detection method based on 3D machine vision is built into a host computer or an industrial computer; the hardware part is set in a connector multi-pin self-steady-state detection device based on 3D machine vision; the host computer or the industrial computer works in coordination with the connector multi-pin self-steady-state detection device based on 3D machine vision to jointly complete the detection of the products to be tested on the connector product production line one by one.

[0008] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention provides a method for improving the stability of each group of pins in the process of calculation and analysis through the coordinated cooperation of software and hardware, especially after two calculations of the center of gravity distance, which simultaneously improves the fit, compatibility, stability and accuracy. While improving the detection accuracy and efficiency, it also solves the problem of unstable yield in the production process. The present invention specifically integrates point cloud data processing and three-dimensional image modeling based on 3D image processing through methods such as pin stability calculation, processes and analyzes each pin of the connector, and applies a variety of data processing technologies to multiple control steps, so that the stability of each group of pins in the process of calculation and analysis has stronger fit, compatibility, stability and accuracy, which can better solve the problem of unstable yield in the production process, can significantly improve the quality and efficiency of production enterprises, and save a lot of conventional detection costs; the detection efficiency is high, the defect types are many, and the application is wide.

[0009] 2. The present invention realizes the coordination of software and hardware, and is convenient and fast in actual use. The detection device adopts a table-top design and can be placed at any measuring station on the production line as needed. The measurement can be started when the power is turned on. When measuring different connector products, only a universal stage needs to be selected, and there is no need to process various special fixtures. The lower surface of the product to be tested on the production line can be quickly measured comprehensively. There are many types of measured data, including weld foot flatness, height, product deformation, etc. One machine can be used for multiple purposes, which can greatly improve production efficiency.

[0010] 3. The improved method and system of the present invention comprehensively weigh multiple factors: the algorithm formula with multiple steps adopts a comprehensive approach to the three major stability-related parameters of triangle area, centroid distance, and height difference, and through weights and nonlinear transformations, finely balances their relative importance and actual contributions, and systematizes the multi-factor evaluation model of area physical stability, centroid distance geometric stability, and height difference structural error to improve detection efficiency and simultaneously meet the needs of detecting different types of defects.

[0011] 4. The improved method and system of the present invention introduces normalization and standardization processing, normalizes the area, center of gravity distance and height difference respectively, ensures that each parameter is evaluated under the same dimension, avoids deviations caused by inconsistent dimensions, ensures that the score is universal and independent of the scale of the input data, and provides an intuitive score range (0-100).

[0012] 5. The present invention introduces nonlinear weight modeling technology and uses square, square root, and exponential functions to model the influence of different parameters, so that the formula can impose higher penalties on combinations with poor stability, while being sensitive to combinations close to the ideal state, amplifying the impact of the penalty on the square of the height difference, emphasizing the destructiveness of large errors to stability, smoothing the situation with a small center of gravity distance to avoid excessive punishment, and imposing exponentially strong penalties on combinations far from the geometric center, thereby improving the robustness of the formula and improving detection efficiency.

[0013] 6. The present invention introduces segmented processing and nonlinear fusion technology, and the segmented processing of the height difference term avoids the misleading influence of negative height difference on the formula. Only the positive height difference is evaluated, making the penalty mechanism more reasonable. The piecewise function based on physical significance combined with the nonlinear model can more accurately characterize the influence of the hyperplane error in the support system and improve the accuracy of the detection results.

[0014] 7. The present invention performs full score normalization and improves interpretability. The multiple algorithm formulas it adopts directly include the step of normalization to 0-100. By setting the theoretical maximum score as the area weight, an easy-to-understand scoring range is provided, and the calculation results of multiple formulas are mapped to a standardized scoring system to meet the intuitive requirements in engineering evaluation or optimization scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 : A schematic diagram of the module composition structure of a connector multi-pin self-steady-state detection system based on 3D machine vision according to an embodiment of the present invention; Figure 2 : is a flow chart of a method for detecting multiple pins of a connector in a self-steady state based on 3D machine vision according to an embodiment of the present invention; Figure 3 : is a schematic diagram of the overall three-dimensional appearance structure of a connector multi-pin self-steady-state detection device based on 3D machine vision according to an embodiment of the present invention; Figure 4 : is a schematic diagram of the internal structure of the front side of the connector multi-pin self-steady-state detection device based on 3D machine vision according to an embodiment of the present invention; Figure 5 : is a schematic diagram of the internal structure of the back side of the connector multi-pin self-steady-state detection device based on 3D machine vision according to an embodiment of the present invention; Figure 6 : It is a schematic diagram of the internal structure of the connector multi-pin self-steady-state detection device based on 3D machine vision according to an embodiment of the present invention.

[0016] In the figure: 1. Body base; 2. Fixed support frame; 3. 3D module mounting plate; 4. Stage support base; 5. X-axis moving module; 6. Mobile drive; 7. Stage; 8. 3D scanning head; 9. Light mobile protection chain; 10. 3D data collector; 11. Power adapter; 12. Vision controller; 13. Start button; 14. Power switch; 15. Alarm; 16. Box. DETAILED DESCRIPTION

[0017] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0018] Example See attached Figure 1-6 The present invention provides a connector multi-pin self-steady-state detection method based on 3D machine vision. It first automatically calculates the position information and height of each pin according to the data collected by the 3D camera, simulates the stable support posture of the connector product on the plane through an algorithm, constructs the three most stable points of the plane support, and then attaches a weight value to each type of calculation parameter, calculates the final score of each three groups of pins, and establishes a reference plane to calculate the data by obtaining the three pins with the highest scores. Finally, it is compared with the product tolerance standard to determine whether the connector product to be tested is qualified. Those that exceed the tolerance range are judged as unqualified products, and those that do not exceed the tolerance range are judged as qualified products.

[0019] The specific steps include: Step 1: Control the 3D scanning head to perform a full range of motion scanning on the surface of the first multi-pin connector product to be tested on the production line. During the scanning process, the laser line cross-sectional profile data is captured and recorded in real time; specifically: Step 1-1: Set up the host computer or industrial computer; Step 1-2: The program implementing steps 1 to 13 is built into the host computer or industrial computer; Step 1-3: Connect the host computer or industrial computer to the connector multi-pin self-steady-state detection device based on 3D machine vision and the connector production line control system respectively; Step 1-4: Deploy the connector multi-pin self-steady-state detection device based on 3D machine vision to the connector product production line; Step 1-5: The host computer or industrial computer works together with the connector multi-pin self-steady-state detection device based on 3D machine vision to detect the products to be tested on the connector product production line one by one; Step 2: The data collector converts the captured laser line cross-sectional profile data into height data according to the principle of laser triangulation, and the converted data forms a series of height profile data sets; Step 3: The height profile data set collected in step 2 is transmitted to the visual controller, which corrects and restores the received data to generate complete three-dimensional point cloud data; Step 3-1: Further perform median filtering and invalid data interpolation processing on the three-dimensional point cloud data generated in step 3 to filter out abnormal data and fill in holes; Step 4: Segment the features of the connector pins in the point cloud data according to the height range, and use the point cloud segmentation algorithm to detect the Z height and corresponding X, Y coordinates of each pin; Step 5: Traverse all the pin data, and every three pins form a combination; Step 6: Use the feature point detection algorithm to calculate the three reference points in each combination, and calculate the area of ​​the triangle formed by the three reference points according to formula 1 Area , and the fractional ratio S1 obtained by calculating the area term afterwards: Formula 1 Where: α is the weight of the area. After the area is normalized, it is directly used as a positive influence. The larger the area, the higher the score. Area It is the area of ​​the triangle formed by the lines connecting the three reference points; S1 is the fractional ratio obtained by calculating the subsequent area term; Step 7: Based on the reference plane constructed in step 6, perform a gravity center distance calculation to calculate the center of the plane formed by each combination of pins and the Euclidean distance between the center of gravity of the product and this plane. Dis tan ce , and calculate the fractional ratio S2 of the centroid distance term according to formula 2: Formula 2 Where: β is the weight, which controls the influence of the distance from the center of gravity on the score; k1 is the amplification factor that adjusts the exponential growth rate; Dis tan ce It is the Euclidean distance between the center of the plane formed by each combined stitch and the center of gravity of the product projected onto this plane; For each combination, the Euclidean distance between the center of the plane formed by the stitches and the projection of the product's center of gravity onto this plane. S2 is not a distance, but a fractional ratio obtained by calculating the distance term of the center of gravity. The distance is smoothed by square root, and then an exponential function is used to amplify the negative impact of larger distances; Step 8: Calculate the sum of the height differences of all the remaining pins outside each group projected onto the plane formed by the pins in this group. Height_Diff, And the fractional ratio S3 obtained by calculating the height difference term according to Formula 3: Formula 3 Where: γ is the weight, which controls the influence of height difference on the score; r is the weight, which controls the influence of height difference on the score. Height_Diff It is the sum of the height differences of all the remaining pins outside each group projected onto the plane formed by the pins in this group; In this embodiment, Height_Diff The sum of the height differences of all the remaining pins outside each group projected onto the plane formed by the pins in this group. Here S3 is not the height difference, but the fractional ratio obtained by calculating the height difference term; A square penalty is applied to positive height differences to reduce the negative impact of small height differences. The part with height difference less than 0 is directly assigned 0 and does not affect the score; Step 9: Using the data calculated in steps 6-8, calculate the stability scores of any three pins that may be used to construct support points, and select the three pins with the highest stability scores as the support points, i.e., the reference points, for subsequent product simulations; Step 10: Use the calculated three reference points to construct the reference plane; Step 11: Recalculate the heights of all pins to the reference plane according to Formula 4, perform secondary gravity center distance calculation, and obtain flatness data S0; Formula 4 In the formula: 100 / α: means to ensure that the theoretical full score is 100, α is the weight of the area, corresponding to the score when the theoretical score is full; Step 12: Compare the calculated data of the first connector product to be tested with the product tolerance standard to determine whether the connector product is qualified. If it exceeds the tolerance range, it will be judged as a defective product, and if it does not exceed the tolerance range, it will be judged as a qualified product. Multiple pin defects of connector products (including crooked pins, broken pins, deformation, etc.) can be detected at one time; Step 13: Repeat steps 1 to 12 to test each multi-pin connector product to be tested on the production line in turn until the production of this batch of multi-pin connector products is completed.

[0020] A connector multi-pin self-steady-state detection system based on 3D machine vision, which includes an image acquisition module, an image processing module, and a data calculation module, wherein the image acquisition module includes an industrial light source and an industrial camera, the image processing module includes a point cloud data processing program and a program for generating 3D pictures; the data calculation module includes an area calculation program, a center of gravity distance calculation program, a secondary center of gravity calculation program, and a score calculation program for each group of pins; The control program software part that implements the aforementioned connector multi-pin self-steady-state detection method based on 3D machine vision (step 1 to step 13) is built into a host computer or an industrial computer; the hardware part is set in a connector multi-pin self-steady-state detection device based on 3D machine vision; the host computer or the industrial computer works in coordination with the connector multi-pin self-steady-state detection device based on 3D machine vision to jointly complete the detection of the products to be tested on the connector product production line one by one.

[0021] The connector multi-pin self-steady-state detection device based on 3D machine vision includes: a body base 1, a fixed support frame 2, and a 3D module mounting plate 3, and is characterized in that: the front of the 3D module mounting plate 3 is provided with a stage support seat 4, an X-axis moving module 5, and a moving driver 6, the stage support seat 4 is provided with a stage 7, the X-axis moving module 5 is provided with a 3D scanning head 8, and the 3D scanning head 8 is connected to the 3D module mounting plate 3 through a light mobile protection chain 9.

[0022] The fixed support frame is vertically installed in the middle of the body base, and the front side of the fixed support frame 2 is fixedly connected to the back side of the 3D module mounting plate 3; a 3D data collector 10, a power adapter 11, a visual controller 12, a start button 13, a power switch 14 and an alarm 15 are provided on the back side of the fixed support frame 2, and the 3D data collector 10, the power adapter 11, the visual controller 12, the start button 13, the power switch 14 and the alarm 15 are electrically connected to the mobile driver 6 and the 3D scanning head 8 respectively; the 3D data collector 10 is arranged at the upper left end of the back side of the fixed support frame 2, the power adapter 11 is arranged at the upper right end of the back side of the fixed support frame 2, and the visual controller 12 is arranged at the lower end of the back side of the fixed support frame 2. A box 16 is also provided on the body base 1, and the box 16 is detachably connected to the body base 1 by screws.

[0023] The image acquisition module in the present invention consists of two parts: an industrial LED light source and an industrial camera. The industrial light source highlights the features of the product being inspected. The industrial camera completes image acquisition and transmits the acquired image to the image processing system using the Gige protocol.

[0024] The image processing system is divided into two parts: 1. 3D point cloud data processing. There is usually noise in the collected data, such as outliers caused by light interference, measurement errors, etc. We use filtering to remove these outliers and use feature extraction, point cloud registration and other methods to obtain the x, y, z coordinates of each feature; 2. Use the processed point cloud data to reconstruct the 3D model of the product and generate a 3D picture.

[0025] The data calculation module calculates each group of stitches by obtaining data such as the stitch position, and calculates the area, center of gravity, height difference and other data of the area formed by each group of stitches. It uses the calculation formula to normalize these parameters and assign different weight values ​​after nonlinear changes, and calculates the stability score of this group of stitches.

[0026] The device of the present invention adopts three-dimensional scanning technology. The surveyor only needs to place the product on a universal stage and press the start button. The three-dimensional scanning module in the instrument will perform a 3D scan on the lower surface of the product. The 3D head collects the contour data and transmits it to the data collector to convert it into height data. Then the scanned series of height data are transmitted to the visual controller for comprehensive processing to form a three-dimensional image. Then the pin features in the image are analyzed and measured to calculate the flatness data of the product; then the product quality is judged according to the measurement tolerance, and the products that do not meet the quality standards are judged as bad products and warned by an alarm sound. The present invention can record all measurement data in a database for easy query and tracing.

[0027] In summary, the embodiments of the present invention adopt software and hardware collaboration, integrate methods such as point cloud data processing based on 3D images and three-dimensional image modeling, to process and analyze connector pins. During the calculation and analysis process of the stability of each group of pins, the fitting, compatibility, stability and accuracy are stronger. At the same time, it also solves the problem of unstable yield in the production process. It has high detection efficiency, targets many types of product defects, and has a wide range of applications.

[0028] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A connector multi-pin self-steady-state detection method based on 3D machine vision, characterized in that: First, the position information and height of each pin are automatically calculated based on the data collected by the 3D camera. The algorithm is used to simulate the stable support posture of the connector product on the plane, and the three most stable points of the plane support are constructed. Then a weight value is attached to each type of calculation parameter, and the final score of each three groups of pins is calculated. The three pins with the highest scores are used to establish a reference plane to calculate the data. Finally, the data is compared with the product tolerance standard to determine whether the connector product to be tested is qualified. Those that exceed the tolerance range will be judged as unqualified products, and those that do not exceed the tolerance range will be judged as qualified products.

2. The connector multi-pin self-steady-state detection method based on 3D machine vision according to claim 1, characterized in that: It includes the following steps: Step 1: Control the 3D scanning head to perform a full range of motion scanning on the surface of the first multi-pin connector product to be tested on the production line. During the scanning process, the laser line cross-sectional profile data is captured and recorded in real time; Step 2: The data collector converts the captured laser line cross-sectional profile data into height data according to the principle of laser triangulation, and the converted data forms a series of height profile data sets; Step 3: The height profile data set collected in step 2 is transmitted to the visual controller, which corrects and restores the received data to generate complete three-dimensional point cloud data; Step 4: Segment the features of the connector pins in the point cloud data according to the height range, and use the point cloud segmentation algorithm to detect the Z height and corresponding X, Y coordinates of each pin; Step 5: Traverse all the pin data, and every three pins form a combination; Step 6: Use the feature point detection algorithm to calculate the three reference points in each combination, and calculate the area of ​​the triangle formed by the three reference points according to formula 1 Area , and the fractional ratio S1 obtained by calculating the area term afterwards: Formula 1 Where: α is the weight of the area. After the area is normalized, it is directly used as a positive influence. The larger the area, the higher the score. Area It is the area of ​​the triangle formed by the lines connecting the three reference points; S1 It is the fractional ratio obtained by calculating the area term afterwards; Step 7: Based on the reference plane constructed in step 6, perform a gravity center distance calculation to calculate the center of the plane formed by each combination of pins and the Euclidean distance between the center of gravity of the product and this plane. Dis tan ce , and calculate the fractional ratio S2 of the centroid distance term according to formula 2: Formula 2 Where: β is the weight, which controls the influence of the distance from the center of gravity on the score; k1 is the amplification factor that adjusts the exponential growth rate; Dis tan ce It is the Euclidean distance between the center of the plane formed by each combined stitch and the center of gravity of the product projected onto this plane; The distance is smoothed by square root, and then an exponential function is used to amplify the negative impact of larger distances; Step 8: Calculate the sum of the height differences of all the remaining pins outside each group projected onto the plane formed by the pins in this group. Height_Diff, And the fractional ratio S3 obtained by calculating the height difference term according to Formula 3: Formula 3 Where: γ is the weight, which controls the influence of height difference on the score; r is is the weight, which controls the influence of height difference on the score; Height_Diff It is the sum of the height differences of all the remaining pins outside each group projected onto the plane formed by the pins of this group of combinations; A square penalty is applied to positive height differences to reduce the negative impact of small height differences. The part with height difference less than 0 is directly assigned 0 and does not affect the score; Step 9: Using the data calculated in steps 6-8, calculate the stability scores of any three pins that may be used to construct support points, and select the three pins with the highest stability scores as the support points, i.e., the reference points, for subsequent product simulations; Step 10: Use the calculated three reference points to construct the reference plane; Step 11: Recalculate the heights of all pins to the reference plane according to Formula 4, perform secondary gravity center distance calculation, and obtain flatness data S0; Formula 4 In the formula: 100 / α: means to ensure that the theoretical full score is 100, α is the weight of the area, corresponding to the score when the theoretical score is full; Step 12: Compare the calculated data of the first connector product to be tested with the product tolerance standard to determine whether the connector product is qualified. If the connector product exceeds the tolerance range, it is determined to be a defective product, and if the connector product does not exceed the tolerance range, it is determined to be a qualified product. Step 13: Repeat steps 1 to 12 to test each multi-pin connector product to be tested on the production line in turn until the production of this batch of multi-pin connector products is completed.

3. The connector multi-pin self-steady-state detection method based on 3D machine vision according to claim 2, characterized in that: Step 1 specifically includes the following steps: Step 1-1: Set up the host computer or industrial computer; Step 1-2: The program implementing steps 1 to 13 is built into the host computer or industrial computer; Step 1-3: Connect the host computer or industrial computer to the connector multi-pin self-steady-state detection device based on 3D machine vision and the connector production line control system respectively; Step 1-4: Deploy the connector multi-pin self-steady-state detection device based on 3D machine vision to the connector product production line; Step 1-5: The host computer or industrial computer works together with the connector multi-pin self-steady-state detection device based on 3D machine vision to detect the products to be tested on the connector product production line one by one.

4. The connector multi-pin self-steady-state detection method based on 3D machine vision according to claim 2, characterized in that: Step 3 also includes the following steps: Step 3-1: Further perform median filtering and invalid data interpolation processing on the three-dimensional point cloud data generated in step 3 to filter out abnormal data and fill in holes.

5. A connector multi-pin self-steady-state detection system based on 3D machine vision, characterized in that: It includes an image acquisition module, an image processing module, and a data calculation module, wherein the image acquisition module includes an industrial light source and an industrial camera, the image processing module includes a point cloud data processing program and a program for generating 3D pictures; the data calculation module includes an area calculation program, a center of gravity distance calculation program, a secondary center of gravity calculation program, and a score calculation program for each group of stitches; The control program software part that implements the connector multi-pin self-steady-state detection method based on 3D machine vision as described in claim 1 is built into a host computer or an industrial computer; the hardware part is arranged in a connector multi-pin self-steady-state detection device based on 3D machine vision; the host computer or the industrial computer works in coordination with the connector multi-pin self-steady-state detection device based on 3D machine vision to jointly complete the detection of the products to be tested on the connector product production line one by one.

6. The connector multi-pin self-steady-state detection system based on 3D machine vision according to claim 5, characterized in that: The connector multi-pin self-steady-state detection device based on 3D machine vision comprises: A machine body base (1), a fixed support frame (2), and a 3D module mounting plate (3); the front of the 3D module mounting plate (3) is provided with a stage support seat (4), an X-axis movable module (5), and a movable driver (6); the stage support seat (4) is provided with a stage (7); the X-axis movable module (5) is provided with a 3D scanning head (8); and the 3D scanning head (8) is connected to the 3D module mounting plate (3) via a light movable protection chain (9).

7. The connector multi-pin self-steady-state detection system based on 3D machine vision according to claim 5, characterized in that: The fixed support frame (2) is vertically mounted in the middle of the machine body base (1), and the front side of the fixed support frame (2) is fixedly connected to the back side of the 3D module mounting plate (3).

8. The connector multi-pin self-steady-state detection system based on 3D machine vision according to claim 5, characterized in that: A 3D data collector (10), a power adapter (11), a visual controller (12), a start button (13), a power switch (14) and an alarm (15) are provided on the back of the fixed support frame (2); the 3D data collector (10), the power adapter (11), the visual controller (12), the start button (13), the power switch (14) and the alarm (15) are electrically connected to the mobile driver (6) and the 3D scanning head (8) respectively.

9. The connector multi-pin self-steady-state detection system based on 3D machine vision according to claim 5, characterized in that: The 3D data collector (10) is arranged at the upper left end of the back side of the fixed support frame (2), the power adapter (11) is arranged at the upper right end of the back side of the fixed support frame (2), and the visual controller (12) is arranged at the lower end of the back side of the fixed support frame (2).

10. The connector multi-pin self-steady-state detection system based on 3D machine vision according to claim 5, characterized in that: A box body (16) is also provided on the machine body base (1), and the box body (16) is detachably connected to the machine body base (1).

Images