Visual positioning method and system based on ROI dynamic alignment

Abstract

The invention discloses a visual positioning method and system based on ROI dynamic alignment, and relates to the technical field of machine vision. The method comprises the following steps: moving a camera to a mark point estimation position; calculating the offset between the mark point and the view center; if the offset is too large, the camera is driven to move to search points according to a specific path (such as a concentric-square-shaped path) in a preset local area; after the mark point is found, the mark point is moved to the view center through mechanical compensation; and finally, high-precision identification is carried out in a central low-distortion area. According to the method, the ROI is dynamically adjusted, so that the mark point is always identified in the optimal imaging area, the edge effect caused by lens edge distortion is fundamentally eliminated, high-precision, high-efficiency and low-cost visual positioning is realized under the condition that a telecentric lens or a complex distortion model is not needed, and the method is suitable for large-scale popularization and application. The method is especially suitable for the alignment process in the high-end precision manufacturing field of semiconductors, display panels and the like.

Description

Technical Field

[0001] This invention relates to the field of machine vision and precision automation equipment technology, specifically to a visual positioning method and system based on ROI dynamic alignment. Background Technology

[0002] In high-end equipment manufacturing fields such as semiconductor packaging and high-precision mounting, vision alignment is a key step in achieving precision machining and assembly. For example, the current mainstream method for semiconductor bonding alignment is to use a fixed camera to photograph the Mark (i.e., the marking point on the wafer or chip), use pattern recognition (PR) algorithms to calculate its center coordinates, and compare them with the target position to guide the mechanical platform to perform compensation alignment.

[0003] The above method has an inherent "edge effect" problem: due to the image distortion that is common in optical lenses, when the marker point is located in the edge area of ​​the camera's field of view, its image will be distorted, resulting in a large error between the center coordinates calculated by the PR algorithm and the actual coordinates, which seriously restricts the overall alignment accuracy of the device.

[0004] To circumvent the above problems, existing technologies typically employ two solutions:

[0005] Option 1: Use an expensive telecentric lens to reduce image distortion, but this will significantly increase the system cost;

[0006] Option 2 involves establishing a complex full-view distortion correction model for real-time correction, but this would significantly increase the computational complexity and calibration time of the system, reducing production efficiency.

[0007] Therefore, there is an urgent need in this field for a new visual positioning solution that can guarantee high alignment accuracy while also possessing the advantages of high efficiency and low cost. Summary of the Invention

[0008] The technical problem to be solved by the present invention is to overcome the shortcomings of the prior art and provide a visual positioning method and system based on ROI dynamic alignment.

[0009] To solve the above-mentioned technical problems, the positioning method in this invention adopts the following technical solution: a visual positioning method based on ROI dynamic alignment, applied to a visual positioning system including a support platform, a visual imaging unit, a driving mechanism, and a control unit. The workpiece requiring alignment is positioned via the support platform. This visual positioning method includes the following steps: S1: The control unit controls the driving mechanism to move the support platform with the positioned wafer to a preset starting position according to pre-set parameters. This starting position is within the field of view of the visual imaging unit; S2: The camera in the visual imaging unit captures an image of the workpiece within the field of view and determines whether there are any marker points on the workpiece within the observation area of ​​the image captured by the camera; if the determination result is "no", proceed to step S3; if the determination result is "yes", proceed to step S4; S3: The control unit controls the driving mechanism to move the support platform with the positioned wafer to a preset starting position. The moving mechanism, starting from the initial position, traverses a path within a local area around the initial position, continuously capturing images with a camera during the movement to find marker points. If a marker point is found within the local area, proceed to step S4; if no marker point is found within the local area, the positioning is deemed a failure, and the preset starting position is reset. S4: Determine the coordinate difference between the center of the marker point and the center of the camera's field of view, and determine if the difference is less than a set threshold. If the difference is greater than or equal to the set threshold, proceed to step S5; if the difference is less than the set threshold, proceed to step S6. S5: If the difference is greater than or equal to the set threshold, the control unit controls the driving mechanism to mechanically compensate for the difference, making the difference less than the set threshold. S6: If the difference is less than the set threshold, the positioning is complete.

[0010] Furthermore, in the above technical solution, in step S3, the path traversal movement adopts a "U" shaped path.

[0011] Furthermore, in the above technical solution, in step S3, the preset local area is a rectangular area with a starting point as the center and a side length of 3-10mm.

[0012] Furthermore, in the above technical solution, the preset local area is a square area of ​​5mm×5mm.

[0013] Furthermore, in the above technical solution, the image observation area refers to the low-distortion area located in the center of the camera's field of view.

[0014] Furthermore, in the above technical solution, after step S5 is completed, the process returns to step 4 to recalculate the coordinate difference between the center of the marker point and the center of the field of view captured by the camera, and then makes a judgment again.

[0015] Furthermore, in the above technical solution, the vision imaging unit includes at least one of a camera located above the carrying platform and a camera located below the carrying platform.

[0016] The positioning system in the present invention adopts the following technical solution: The system includes: a carrying platform for carrying and positioning the workpiece to be aligned; a vision imaging unit including at least one of a camera located above the carrying platform and a camera located below the carrying platform, and the optical path of the camera points to the carrying platform; a driving mechanism connected to the carrying platform for driving the carrying platform to move within a plane; a control unit electrically connected to the driving mechanism; and the system adopts the foregoing positioning method.

[0017] The present invention can be used in the field of high-end equipment manufacturing for various high-precision mounting applications such as wafer bonding, die packaging, and chip bonding. It is a vision positioning method based on dynamic alignment of ROI (Region of Interest), which eliminates the error caused by the "edge effect" fundamentally by dynamically moving the observation area of the vision system, enabling the marked point to be always recognized within the low-distortion area at the center of the camera's field of view. While ensuring high precision, it also has the advantages of high efficiency and low cost.

[0018] After adopting the above technical solution, compared with the prior art, the present invention has the following remarkable advantages:

[0019] 1. By dynamically adjusting, the marked point is always imaged within the low-distortion area at the center of the camera's field of view, fundamentally eliminating the influence of lens edge distortion (edge effect) on the pattern recognition accuracy, and achieving sub-pixel-level high-precision positioning.

[0020] 2. The present invention does not require the use of expensive telecentric lenses, nor does it need to construct and maintain a complex full-field distortion correction model, reducing costs and the complexity of system development and maintenance.

[0021] 3. The present invention adopts a "loop" - shaped point search algorithm, which is efficient and has strong robustness. It can quickly find the marked point within the allowable small deviation range, avoiding unnecessary stoppages and improving the overall production efficiency of the equipment. Description of the Drawings

[0022] Figure 1 is the overall flowchart of the vision positioning method of the present invention;

[0023] Figure 2 is the schematic diagram of the "loop" - shaped point search path in the present invention. Detailed Embodiments

[0024] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0025] The visual positioning method of this invention is applicable to various workpieces such as wafers, dies, and chips, enabling high-precision mounting in high-end equipment manufacturing fields, including wafer bonding, die packaging, and chip assembly. For simplicity, the following description uses wafer alignment in semiconductor wafer bonding processes as an example, but the application of this invention is not limited to this.

[0026] Taking wafer alignment in wafer bonding process as an example, the positioning system of the present invention includes:

[0027] A carrier platform is used to support and position the wafers to be bonded. Typically, a carrier platform includes an upper carrier platform and a lower carrier platform to position the upper and lower wafers to be bonded, respectively.

[0028] A vision imaging unit, comprising at least one camera, is used for image capture during visual alignment. Typically, if the upper wafer is used as a reference, visual positioning of the lower wafer is required. The camera is positioned above the lower support platform and captures images of the lower wafer for visual positioning. Conversely, if the lower wafer is used as a reference, visual positioning of the upper wafer is required. The camera is positioned below the upper support platform and captures images of the upper wafer for visual positioning. Of course, it is common practice to simultaneously use both upper and lower cameras to capture images of the upper and lower wafers respectively.

[0029] A drive mechanism, connected to the support platform (including upper and lower support platforms), is used to drive the support platform to move within a plane. Typically, the drive mechanism can employ a high-precision XY linear motor platform to achieve free movement within the plane.

[0030] The control unit is electrically connected to the drive mechanism. The control software running in the control unit executes the positioning system to perform positioning operations according to a preset process.

[0031] like Figure 1 As shown, based on the above-described visual positioning system, the implementation flow of the method of the present invention is as follows:

[0032] S1: Starting Position

[0033] The control unit, based on preset parameters, controls the drive mechanism to move the wafer-positioned carrier platform to a preset starting position, which is within the field of view of the visual imaging unit. The coordinates of this starting position are within the field of view of the corresponding camera in the visual imaging unit.

[0034] S2: ROI Pre-positioning

[0035] The camera in the visual imaging unit takes images of the wafer within its field of view and determines whether there are marker points on the wafer within the observation area of ​​the image taken by the camera.

[0036] Because images captured by a camera contain distorted areas (edge ​​regions), to avoid calculation errors caused by edge distortion, this invention selects a low-distortion region at the center of the camera's field of view as the image observation area, based on the camera's own parameters. The specific size of this image observation area can be determined by considering factors such as the camera's own parameters and the focal length. Compared to existing fixed-field-of-view recognition methods, this invention only checks the presence of a Mark (marker point) set on the wafer within the "image observation area," thereby reducing calculation errors caused by lens edge distortion. In other words, this image observation area is the optimal recognition area for the Marker point.

[0037] If no marker is observed on the wafer within the image observation area, the result is "No", and proceed to step S3; if a marker is observed on the wafer within the image observation area, the result is "Yes", and proceed to step S4.

[0038] S3: Dynamic Point Finding

[0039] The control unit controls the drive mechanism to traverse a path within a local area around the starting position, starting from the initial position, and continuously captures images with a camera during the movement to find marker points.

[0040] Specifically, if no marker point is observed on the wafer within the image observation area, it means the marker point does not fall within the "image observation area," and the marker point's location needs to be located. During the search, only a local area is searched. The size of this local area should not be too small to avoid unnecessary errors due to minor differences. Of course, if the size of the local area is too large, the point-finding efficiency will decrease, and an error should be reported directly, indicating that the preset starting position has a large deviation and needs to be reset. Typically, the preset local area is a rectangular area centered on the starting point with a side length of 3-10mm. Preferably, the preset local area is a 5mm × 5mm square area.

[0041] Furthermore, to ensure that all locations within a local area are queried, this invention employs a "loop"-shaped path traversal and moving sampling method. Combined with... Figure 2 As shown, taking a square area of 5mm×5mm as the preset local area as an example, the control unit controls the driving mechanism to gradually move in a "return" - shaped path from the inside to the outside with this starting point position as the center. For each movement of a step length (e.g., 0.5mm), the camera takes an image and determines whether there are marking points within the image observation area. Once a marking point is found at a certain position, the traversing movement stops and step S4 is entered.

[0042] Of course, it is also possible that after traversing the entire preset local area, no marking point on the wafer is found. Then, it is determined that the current positioning fails, and an error is directly reported, indicating that there is a large deviation in the preset starting position, and the preset starting position needs to be reset.

[0043] S4: Deviation judgment

[0044] Through step S2 or S3, a marking point on the wafer is observed within the image observation area. At this time, it is necessary to determine the coordinate difference between the center of the marking point and the center of the camera's shooting field of view, and judge whether this difference is less than the set threshold.

[0045] For example, the center coordinates of the found marking point are (X0, Y0), and the coordinates of the center of the field of view are (X1, Y1). Calculate the difference (Δx, Δy) between the two coordinates. Judge whether (Δx, Δy) is within the allowed threshold range. If the difference (Δx,Δy) is greater than or equal to the set threshold (e.g., a certain number of pixel values), step S5 is entered. If the difference (Δx, Δy) is less than the set threshold, step S6 is entered.

[0046] S5: Mechanical compensation

[0047] If the difference (Δx, Δy) is greater than or equal to the set threshold, the control unit controls the driving mechanism to perform mechanical compensation on the difference to make the difference less than the set threshold.

[0048] To ensure the accuracy of mechanical compensation, after the compensation is completed, it is necessary to return to step 4 again, recalculate the coordinate difference between the center of the marking point and the center of the camera's shooting field of view, and make a judgment again to ensure the accuracy of the compensation.

[0049] S6: If the difference is less than the set threshold, the positioning is completed. At this time, the center coordinates of the current marking point coincide with the coordinates of the field center, the wafer is aligned, and the positioning before bonding is achieved.

[0050] The visual imaging unit includes at least one of a camera located above the bearing platform and a camera located below the bearing platform. [[ID=二十九]] [[ID=三十]]

[0051] For wafer bonding processes, the above process only describes the visual positioning of one wafer. In actual use, positioning operations need to be performed separately for the upper and lower wafers.

[0052] Applying the method of this invention in actual wafer bonding equipment, a significant improvement in key performance indicators can be observed by comparing positioning data before and after enabling the ROI dynamic alignment function. For example, the positioning repeatability accuracy (3σ value) can be reduced from several micrometers to the submicrometer level; the process capability index (CPK), which reflects the overall process capability, can be improved from less than 1.33 to more than 1.67, proving that the method can stably produce high-precision alignment results and greatly improve the bonding yield.

[0053] It should be noted that the above embodiments use a mobile support platform (workpiece) as an example. Following the principles of this invention, a mobile camera or a camera-platform coordinated movement can also be used to achieve the "relative movement between the camera and the marker point." The described "U"-shaped path and 5mm×5mm search area are preferred embodiments, and can be reasonably adjusted according to camera field of view, accuracy requirements, and efficiency in practical applications.

[0054] Of course, the above description is only a specific embodiment of the present invention and is not intended to limit the scope of the present invention. All equivalent changes or modifications made to the structure, features and principles described in the claims of the present invention should be included in the scope of the claims of the present invention.

Claims

1. A visual positioning method based on ROI dynamic alignment, applied to a visual positioning system including a support platform, a visual imaging unit, a drive mechanism, and a control unit, wherein the workpiece to be aligned is positioned by the support platform, characterized in that: The visual positioning method includes the following steps: S1: The control unit controls the drive mechanism to move the carrying platform with the workpiece positioned to a preset starting position according to the original preset parameters. The starting position is within the field of view of the vision imaging unit. S2: The camera in the visual imaging unit takes images of the workpiece within its field of view and determines whether there are any markers on the workpiece within the observation area of ​​the image taken by the camera. If the judgment result is "no", then proceed to step S3. If the judgment result is "yes", then proceed to step S4; S3: The control unit controls the drive mechanism to move along a path within a local area around the starting position, starting from the starting position, and continuously captures images with a camera during the movement to find marker points; If a marker point is found within the local area, proceed to step S4. If no marker point is found in the local area, the positioning is deemed to have failed, and the preset starting position is reset. S4: Determine the coordinate difference between the center of the marker point and the center of the field of view captured by the camera, and determine whether the difference is less than a set threshold. If the difference is greater than or equal to the set threshold, proceed to step S5. If the difference is less than the set threshold, proceed to step S6; S5: If the difference is greater than or equal to the set threshold, the control unit controls the drive mechanism to mechanically compensate for the difference, so that the difference is less than the set threshold. S6: If the difference is less than the set threshold, the positioning is completed.

2. The visual positioning method based on ROI dynamic alignment according to claim 1, characterized in that: In step S3, the path traversal movement adopts a "back" shaped path.

3. The visual positioning method based on ROI dynamic alignment according to claim 1, characterized in that: In step S3, the preset local area is a rectangular area with a starting point as the center and a side length of 3-10mm.

4. The visual positioning method based on ROI dynamic alignment according to claim 3, characterized in that: The preset local area is a square area of ​​5mm × 5mm.

5. The visual positioning method based on ROI dynamic alignment according to claim 1, characterized in that: The image observation area refers to the low-distortion region located in the center of the camera's field of view.

6. The visual positioning method based on ROI dynamic alignment according to claim 1, characterized in that: After step S5 is completed, return to step 4 to recalculate the coordinate difference between the center of the marker point and the center of the field of view captured by the camera, and make a judgment again.

7. The visual positioning method based on ROI dynamic alignment according to claim 1, characterized in that: The visual imaging unit includes at least one of a camera located above the support platform and a camera located below the support platform.

8. A system for a visual positioning method based on ROI dynamic alignment, characterized in that, The system includes: A support platform is used to support and position the workpieces that need to be aligned. The visual imaging unit includes at least one of a camera located above the support platform and a camera located below the support platform, and the added optical paths point towards the support platform; A drive mechanism, connected to the support platform, is used to drive the support platform to move in a plane; The control unit is electrically connected to the drive mechanism; The system employs a visual positioning method based on ROI dynamic alignment as described in any one of claims 1-6.

Images