A method of creating a probe card based on point cloud data

Abstract

The application discloses a kind of based on point cloud data creation probe board method, including element to be measured, scanning device and PLC, PLC is configured with PCL, and the point cloud data of X series, Y series, Nx series and Ny series are collected by high-definition monitoring camera set on x-y axis, after point cloud completion, point cloud data preprocessing, point cloud data segmentation, point cloud data detection and the relevant procedures of point cloud data registration, the pose information of contact point on the same coordinate plane of the coordinate information it collects is confirmed substrate, the point cloud data of Z series and Nz series are collected by high-definition monitoring camera set on z axis, so that the pose information of probe of the coordinate information it collects is confirmed to the same coordinate plane, and probe board creation is completed, not only can realize the accuracy of measurement data, while also can improve work efficiency.

Description

Technical Field

[0001] This invention belongs to the field of semiconductor testing equipment technology, and specifically relates to a method for creating a probe board based on point cloud data. Background Technology

[0002] Electrical components are crucial parts of electronic products. Before leaving the factory, all electrical components must undergo performance testing to identify quality issues promptly. Current testing methods involve the component under test (DUT) contacting a probe plate, and the resulting data is fed back to an information processing unit for further processing.

[0003] However, in the process of implementing the inventive technical solution in the embodiments of this application, the inventors of this application discovered that the above-mentioned technology has at least the following technical problems: In actual production, the components of the device under test (DUT) are usually assembled. Then, a probe board is produced to match the DUT's contact points or terminal distribution. The probe board consists of a substrate and probes mounted on the substrate. Designers usually use calipers to measure the distance between contact points and the height of the probes. However, due to objective factors such as light or vision, the measurement data can easily become inaccurate, leading to a mismatch between the probe board and the DUT. This requires repeated design and measurement, which is time-consuming, laborious, and inefficient. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this application provides a method for creating a probe board based on point cloud data. This method solves the technical problem in existing technologies where inaccurate measurement data is caused by objective factors, leading to mismatch between the probe board and the component under test, requiring redesign and measurement. This method is highly practical, can achieve accurate measurement data, and improves work efficiency.

[0005] This application provides a method for creating a probe board based on point cloud data, including a device under test, a scanning device, and a PLC (Programmable Logic Controller), wherein the PLC is configured with a PCL (Point Cloud Library). The method includes the following steps: Step a: The scanning device scans the contour information of the cross-section of the component under test with contact points to acquire point cloud data; Step b: Recreate the scene using ContextCapture software and compare the contour information with the component under test to confirm whether the contour information is clear or complete; Step c: If the contour information is unclear, add control points and edit connection points to handle scenes with uneven resolution. If the contour information is incomplete, point cloud completion processing can be performed based on the shape, structure, and topological features of the point cloud data using Poisson reconstruction or marching cubes to restore the missing parts of the data. Step d: If the contour information is clear and complete, preprocess the point cloud data obtained in step a using PCL; Step e: Segment the preprocessed point cloud data using a feature-based method; Step f: Inspect the point cloud data segmented in step e; Step g: Perform point cloud registration on the point cloud data detected in step f; Step h: Extract the point cloud data from step g, select the X series, Nx series, Y series, and Ny series point cloud data and aggregate them into the same coordinate plane to confirm the pose information of the contact points on the substrate, and select the Z series and Nz series point cloud data and aggregate them into the same coordinate plane to confirm the pose information of the probe. The substrate and the probe constitute a probe plate for detecting the device under test.

[0006] Preferably, the scanning device disclosed in this application embodiment includes a stereo vision mechanism and structured light. The stereo vision mechanism is an RGB-D camera. The RGB-D camera includes a high-definition monitoring camera arranged along the xy axis, which is placed directly above the object under test and is used to detect the horizontal position and orientation of the object along the xy axis. The high-definition monitoring camera arranged along the z axis is placed at a vertical position on the side of the object under test and is used to detect the position of the object in the z-axis direction.

[0007] Preferably, the preprocessing described in step d in this application also discloses that the preprocessing includes removing noise from the point cloud data, using Gaussian filtering to remove noise caused by weak reflection signals due to the surface material of the object or the lighting environment, and using mean filtering to remove noise caused by systematic and random errors due to accuracy or resolution of the stereo vision mechanism.

[0008] Preferably, the embodiment of this application also discloses that the feature-based method in step e includes segmentation using local feature information, surface texture information, and color information.

[0009] Preferably, in step f, the detection includes point cloud output preprocessing and three-dimensional target detection and classification. The three-dimensional target detection and classification is based on a 3D framework method to directly establish a 3D framework in space to locate and classify target objects.

[0010] Preferably, the embodiments of this application also disclose that the point cloud registration in step g includes feature extraction, point cloud matching, change estimation and point cloud transformation. The point cloud matching uses the nearest neighbor search method to find corresponding point pairs in two point cloud data and performs matching based on local description similarity. The point cloud transformation is to transform one point cloud data to the coordinate system of the other point cloud data so that the two point cloud data are aligned in the same coordinate system.

[0011] Preferably, this application embodiment also discloses that the structured light illuminates the surface of the target object through a structured light source such as a projection grating or coded light strips, and uses the information of the camera's reflection of the grating or the deformation of the light strips to calculate the point cloud data of each point on the surface of the target object.

[0012] Preferably, the embodiments of this application also disclose that the representation format of the electric cloud data includes PLY format, OBJ format, and STL format.

[0013] The technical solutions provided in this application have at least the following significant technical effects: (1) High-definition monitoring cameras set along the x and y axes collect point cloud data of the X series, Y series, Nx series and Ny series. After point cloud completion, point cloud data preprocessing, point cloud data segmentation, point cloud data detection and point cloud data registration, the coordinate information collected is aggregated into the same coordinate plane to confirm the pose information of the contact point on the substrate. High-definition monitoring cameras set along the z axis collect point cloud data of the Z series and Nz series. The coordinate information collected is aggregated into the same coordinate plane to confirm the pose information of the probe. The probe board is created, which not only ensures the accuracy of the measurement data, but also improves the work efficiency.

[0014] (2) If the surface of the component under test is smooth, high-precision point cloud data can be obtained by structured light based on the reflection characteristics of smooth surface. Attached Figure Description

[0015] Figure 1 This is a flowchart of the method in an embodiment of this application.

[0016] Figure 2 This is a coordinate graph of point cloud data in XYZ format in the embodiments of this application.

[0017] Figure 3 This is a schematic diagram of the cross-section of the element under test with contact points in an embodiment of this application. Implementation

[0018] This application provides a method for creating a probe board based on point cloud data, which solves the technical problem in the prior art where inaccurate measurement data is caused by objective factors, leading to mismatch between the probe board and the component under test, requiring redesign and measurement.

[0019] The technical solution in this application aims to solve the above problems in the following general way: High-definition monitoring cameras set along the x and y axes collect point cloud data from the X, Y, Nx, and Ny series. After point cloud completion, point cloud data preprocessing, point cloud data segmentation, point cloud data detection, and point cloud data registration, the data is aggregated into the same coordinate plane to confirm the pose information of the contact points on the substrate. High-definition monitoring cameras set along the z axis collect point cloud data from the Z and Nz series, which are then aggregated into the same coordinate plane to confirm the pose information of the probes and create a probe board.

[0020] like Figures 1-3 As shown in the figure, this application provides a method for creating a probe board based on point cloud data, including a device under test, a scanning device, and a PLC (Programmable Logic Controller). The PLC is configured with a PCL (Point Cloud Library). The method includes the following steps: Step a: The scanning device scans the contour information of the cross-section of the component under test with contact points to acquire point cloud data. The scanning device includes a stereo vision mechanism and structured light. The stereo vision mechanism is an RGB-D camera. The RGB-D camera includes a high-definition monitoring camera set along the xy-axis, which is placed directly above the component under test to detect the horizontal position and orientation of the component under test along the xy-axis. A high-definition monitoring camera set along the z-axis is placed vertically to the side of the component under test to detect the position of the component under test in the z-axis direction. The structured light illuminates the surface of the target object through a projection grating or an coded light stripe structured light source. The point cloud data of each point on the surface of the component under test is calculated using the information of reflection of the grating or deformation of the light stripe by the camera.

[0021] Step b: Recreate the scene using ContextCapture software to compare the contour information with the component under test to confirm whether the contour information is clear or complete.

[0022] Step c: If the contour information is unclear, add control points and edit connection points to handle scenes with uneven resolution. If the contour information is incomplete, point cloud completion processing can be performed based on the shape, structure and topological features of the point cloud data, using Poisson reconstruction or marching cubes to restore the missing parts of the data.

[0023] Step d: If the contour information is clear and complete, the point cloud data obtained in step a is preprocessed using PCL. The preprocessing includes removing noise from the point cloud data, using Gaussian filtering to remove noise caused by weak reflection signals due to the surface material or lighting environment of the object, and using mean filtering to remove noise caused by systematic and random errors due to accuracy or resolution of the stereo vision mechanism.

[0024] Step e: Segment the preprocessed point cloud data using a feature-based method; the feature-based method includes segmentation based on local feature information, surface texture information, and color information.

[0025] Step f: Detect the point cloud data segmented in step e; this detection includes point cloud output preprocessing and 3D target detection and classification. 3D target detection and classification is based on the 3D framework method, which directly establishes a 3D framework in space to locate and classify target objects.

[0026] Step g: Perform point cloud registration on the point cloud data detected in step f; Point cloud registration includes feature extraction, point cloud matching, change estimation, and point cloud transformation. Point cloud matching uses nearest neighbor search to find corresponding point pairs in two point cloud data, and matching is performed based on local description similarity. Point cloud transformation is to transform one point cloud data to the coordinate system of the other point cloud data so that the two point cloud data are aligned in the same coordinate system.

[0027] Step h: Extract the point cloud data from step g, select the coordinate information of the X series, Nx series, Y series, and Ny series and aggregate them into the same coordinate plane to confirm the pose information of the contact points on the substrate, and select the coordinate information of the Z series and Nz series and aggregate them into the same coordinate plane to confirm the pose information of the probe. The substrate and the probe constitute a probe board for detecting the device under test. The point cloud data can also be represented in PLY format, OBJ format, and STL format.

[0028] Work process: The component under test is moved to the scanning area of ​​the stereo vision mechanism. High-definition monitoring cameras set along the x and y axes collect point cloud data of the X, Y, Nx, and Ny series. After point cloud completion, point cloud data preprocessing, point cloud data segmentation, point cloud data detection, and point cloud data registration, the collected coordinate information is aggregated into the same coordinate plane to confirm the pose information of the contact point on the substrate. High-definition monitoring cameras set along the z axis collect point cloud data of the Z and Nz series, and the collected coordinate information is aggregated into the same coordinate plane to confirm the pose information of the probe. The probe board is then created.

[0029] Although preferred embodiments of the invention have been described, those skilled in the art, once they learn the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications that fall within the scope of the invention.

[0030] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from the spirit and scope of this invention. Therefore, if these modifications and variations of this invention fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for creating a probe board based on point cloud data, wherein, The method includes the device under test, a scanning device, and a PLC, wherein the PLC is configured with a PCL. The method includes the following steps: Step a: The scanning device scans the contour information of the cross-section of the component under test with contact points to acquire point cloud data; Step b: Recreate the scene using ContextCapture software and compare the contour information with the component under test to confirm whether the contour information is clear or complete; Step c: If the contour information is unclear, add control points and edit connection points to handle scenes with uneven resolution. If the contour information is incomplete, point cloud completion processing can be performed using Poisson or marching cubes based on the shape, structure, and topological features of the point cloud data to restore the missing parts of the data. Step d: If the contour information is clear and complete, preprocess the point cloud data obtained in step a using PCL; Step e: Segment the preprocessed point cloud data using a feature-based method; Step f: Inspect the point cloud data segmented in step e; Step g: Perform point cloud registration on the point cloud data detected in step f; Step h: Extract the point cloud data from step g, select X-series, Nx-series, Y-series, and Ny-series coordinate information and aggregate them into the same coordinate plane to confirm the pose information of the contact points on the substrate, and select Z-series and Nz-series coordinate information and aggregate them into the same coordinate plane to confirm the pose information of the probe; The point cloud registration in step g includes feature extraction, point cloud matching, change estimation, and point cloud transformation. The point cloud matching uses nearest neighbor search to find corresponding point pairs in two point cloud data and performs matching based on local description similarity. The point cloud transformation is to transform one point cloud data to the coordinate system of the other point cloud data so that the two point cloud data are aligned in the same coordinate system.

2. The method for creating a probe board based on point cloud data according to claim 1, wherein, The scanning device in step a includes a stereo vision mechanism and structured light. The stereo vision mechanism is an RGB-D camera. The RGB-D camera includes a high-definition monitoring camera set along the xy-axis, which is placed directly above the object under test and is used to detect the horizontal position and orientation of the object along the xy-axis. The high-definition monitoring camera set along the z-axis is placed at a vertical position on the side of the object under test and is used to detect the position of the object in the z-axis direction.

3. The method for creating a probe board based on point cloud data according to claim 1, wherein, The preprocessing described in step d includes removing noise from the point cloud data, using Gaussian filtering to remove noise caused by weak reflection signals due to the surface material of the object or the lighting environment, and using mean filtering to remove noise caused by systematic and random errors due to accuracy or resolution of the stereo vision mechanism.

4. The method for creating a probe board based on point cloud data according to claim 1, wherein, The feature-based method described in step e includes segmentation using local feature information, surface texture information, and color information.

5. The method for creating a probe board based on point cloud data according to claim 1, wherein, In step f, the detection includes point cloud output preprocessing and 3D target detection and classification. The 3D target detection and classification is based on a 3D framework method that directly establishes a 3D framework in space to locate and classify target objects.

6. The method for creating a probe board based on point cloud data according to claim 2, wherein, The structured light illuminates the surface of the target object through a projection grating or an coded light stripe structured light source. The point cloud data of each point on the surface of the target object is calculated using information from the camera's reflection of the grating or the deformation of the light stripe.

7. The method for creating a probe board based on point cloud data according to claim 1, wherein, The point cloud data can also be represented in PLY, OBJ, and STL formats.

Images