Method for infinitely expanding visual field of camera of chip mounter for error compensation

Abstract

The invention discloses a method for infinitely expanding the visual field of a camera of a chip mounter for error compensation, and relates to the technical field of chip mounting. The method for infinitely expanding the visual field of the camera of the chip mounter for error compensation comprises the following steps: S1, establishing a coordinate system and extracting feature points; s2, homogeneous coordinate matrix calculation and coordinate transformation; s3, expanding the visual field; s4, error analysis and compensation; s5, carrying out surface mounting operation; by infinitely expanding the virtual field of view of the camera, the traditional tedious switching process which is easy to introduce errors is avoided. Firstly, a double-coordinate system is defined, feature points are extracted, a homogeneous coordinate matrix is established to complete coordinate transformation and distortion correction, and then new feature points are continuously selected to achieve infinite extension of the visual field so as to cope with a large-range chip substrate. And then analyzing the mounting position deviation and judging the error property, performing error compensation by adopting a method of recalibrating H matrix parameters or measuring for multiple times and averaging, and finally outputting the compensated coordinates to a chip mounter mobile module to complete mounting.

Description

Technical Field

[0001] The present invention relates to the technical field of chip mounting, and particularly to a method for compensating errors by infinitely expanding the camera field of view of a mounter. Background Art

[0002] In the current field of precision electronics manufacturing, as a core process link, chip mounting has almost stringent requirements for mounting accuracy. In order to achieve high-precision chip mounting and positioning, the existing chip mounters usually design the camera field of view to be relatively narrow. This is because in a small field of view scenario, the camera can more focusedly image the features of the chip and the substrate within a specific area. With high-resolution imaging and refined image processing algorithms, it can effectively reduce the positioning deviation caused by imaging blur, edge distortion, etc., so as to achieve extremely high mounting accuracy indicators within a limited space range, meeting the manufacturing requirements of many current high-integration, micro-miniaturized electronic products with extremely low tolerance for chip mounting position errors.

[0003] However, with the increasing complexity of electronic product functions and the continuous expansion of scale, the size of the chip substrate and the range of mounting layouts have also increased significantly. When the position of the chip to be mounted exceeds the current limited field of view of the camera, the conventional calculation method reveals obvious limitations. In actual mounting operations, since the chip substrate coordinate system, the camera coordinate system, and the mechanical module coordinate system are established based on different physical benchmarks and logical rules, they each have independent origins, axial definitions, and unit scales. In the conventional method, in order to accurately transfer the position information of the chip beyond the field of view to the mechanical module to complete the mounting action, it is necessary to frequently switch back and forth between these three coordinate systems for calculation.

[0004] In the process of converting from the chip substrate coordinate system to the camera coordinate system, first, it is necessary to perform a preliminary projection transformation on the coordinates of the target chip on the chip substrate according to the pre-calibrated spatial relationship model. This process involves complex matrix operations and has extremely high requirements for the accuracy of calibration parameters. Any minor calibration error will be amplified in the calculation. Then, after the camera completes the acquisition of chip information within a partial field of view, it is necessary to convert the chip position information obtained under the camera coordinate system to the mechanical module coordinate system so that the actuator such as the robotic arm can accurately position and grasp the chip. During this period, not only geometric transformations such as translation and rotation between coordinate systems need to be considered, but also problems such as scale differences and unit conversions that may exist under different coordinate systems need to be addressed.

[0005] This calculation method of switching back and forth between multiple coordinate systems may introduce new error sources in each conversion step. Calibration errors, rounding errors of coordinate transformation algorithms, dynamic errors during the movement of mechanical modules, etc. accumulate and stack up during multiple coordinate system switches, ultimately resulting in a large deviation between the actual chip mounting position and the theoretically designed position, seriously affecting the quality of chip mounting and the product yield rate, increasing the rework cost and time cost in the production process, and restricting the further development of the chip mounting process towards larger scale and higher precision. Summary of the Invention

[0006] In view of the deficiencies of the prior art, the present invention provides a method for compensating errors by infinitely expanding the camera field of view of a mounter, which solves the problems.

[0007] To achieve the above objectives, the present invention is realized through the following technical solutions: A method for compensating errors by infinitely expanding the camera field of view of a mounter includes the following steps: S1. Define the chip substrate coordinate system and the camera coordinate system, and use the SIFT algorithm to automatically extract the feature points with scale and rotation invariance in the chip substrate image, and obtain the coordinates of the feature points in the two coordinate systems; S2. Establish a homogeneous coordinate matrix H to describe the transformation relationship from the chip substrate coordinate system to the camera coordinate system, transform the points in the chip substrate coordinate system to the camera coordinate system through the matrix H, and consider the camera optical distortion. Adopt a polynomial distortion model to correct the transformed coordinates to eliminate the distortion effect; S3. Continuously select new feature points of the chip substrate coordinate system, repeat the coordinate transformation process, and realize the infinite expansion of the camera field of view to process the chip substrate beyond the actual field of view of the camera; S4. Analyze the deviation between the actual position and the theoretical position of the chip after mounting, and judge the nature of the error; for systematic errors, recalibrate the parameters of the H matrix; for random errors, adopt the method of taking the average of multiple measurements; apply the error compensation strategy to the coordinate transformation process and correct the transformed coordinates; S5. Output the error-compensated camera coordinates to the mounter moving module for chip mounting work.

[0008] Preferably, the S1 specifically includes coordinate system establishment and feature point selection: S1.1 Establish a coordinate system: Define the chip substrate coordinate system , the origin O b is selected at one of the corner points of the chip substrate, and the X b axis and the Y b axis are respectively along two sides of the chip substrate; Define the camera coordinate system , the origin O c is the optical center of the camera, and the Xc The X-axis and the Y-axis c are respectively parallel to the horizontal and vertical directions of the camera imaging plane; Feature point selection: Use the SIFT feature point detection algorithm to process the chip substrate image, and automatically extract feature points with scale invariance and rotation invariance; Suppose n feature points are detected from the chip substrate image, and their coordinates in the chip substrate coordinate system are expressed as , and the corresponding coordinates in the camera coordinate system are expressed as .

[0009] Preferably, the homogeneous coordinate matrix H in S2 is: ; where λ is the scaling factor, representing the scale transformation ratio from the chip substrate coordinate system to the camera coordinate system; θ is the rotation angle, representing the rotation relationship between the two coordinate systems; and are respectively the translation amounts of the feature point in the X and Y directions of the two coordinate systems.

[0010] Preferably, S3 specifically includes coordinate transformation and field of view expansion: S3.1 Coordinate transformation: For any point in the chip substrate coordinate system, its homogeneous coordinate is expressed as , and it is transformed to the coordinate in the camera coordinate system through the homogeneous coordinate matrix H as: ; Considering the optical distortion of the camera, a polynomial distortion model is used for correction; suppose the distortion model is: ; ; where , k1, k2, k3... are distortion coefficients, which are determined through calibration experiments; S3.2 Field of view expansion: Continuously select new feature points in the chip substrate coordinate system and repeat the above coordinate transformation process to achieve infinite expansion of the camera field of view.

[0011] Preferably, the error analysis steps in S4 include: S4.1 Analyze the error source by actually measuring the deviation between the actual position and the theoretical position of the mounted chip; suppose the theoretical position is , and the actually measured position is , then the error , ; S4.2 Analyze the nature of the error to determine whether it is a systematic error or a random error; systematic errors are regular, while random errors are random.

[0012] Preferably, the error compensation strategy in S4 includes: For systematic errors, if a fixed deviation related to the H matrix is found, correct it by recalibrating the parameters in the H matrix; assume the deviation of the translation amount is ΔT x and ΔT y , then the corrected homogeneous coordinate matrix is: ; For random errors, use the method of taking the average of multiple measurements and data smoothing to reduce the impact; Error compensation implementation: Apply the formulated error compensation strategy to the coordinate transformation process to correct the transformed coordinates.

[0013] Preferably, S5 specifically includes: S5.1 Coordinate output: Output the coordinates (x c ′, y c ′) in the camera coordinate system after error compensation and distortion processing to the movement module of the mounter; S5.2 Moving and mounting: The mounter controls the movement module to accurately mount the chip to the specified position on the substrate according to the received coordinate information.

[0014] Preferably, in the analysis of the nature of the error, the analysis of the error regularity includes: S4.2.1 Data collection: Multiple measurements: Under the same experimental conditions, measure the same mounting position m times, and record the actual position coordinates of each measurement ; S4.2.2 Analyze the error regularity: Calculate the mean and variance of the error: Mean: ; ; Variance: ; ; Draw an error distribution diagram: Take the number of measurements i as the horizontal axis and the error and as the vertical axis to draw a graph of the change of the error with the number of measurements, or draw a histogram or scatter plot of the error to observe the distribution of the error; Calculate the autocorrelation coefficient: The autocorrelation coefficient Used to reflect the correlation of errors between different time points or measurement times: ; Wherein, represents the covariance of Δx i and , represents standard deviation of, a represents the lag order.

[0015] Preferably, the systematic error judgment includes: If the error mean or is significantly non-zero, and the error distribution diagram shows an obvious trend or periodic change of the error, or the autocorrelation coefficient is significantly non-zero at a certain lag order a, it indicates the existence of a systematic error.

[0016] Preferably, the random error judgment includes: If the error mean or is close to zero, and the error distribution diagram shows a random distribution of the error, without an obvious trend or periodic change, and at the same time the autocorrelation coefficient is close to zero at all lag orders k, it indicates the existence of a random error.

[0017] The present invention provides a method for compensating errors by infinitely expanding the vision of the camera of a mounter. Compared with the prior art, it has the following beneficial effects: 1. For the method of compensating errors by infinitely expanding the vision of the camera of the mounter, in the conventional method, due to the small vision of the camera, when processing the chip mounting position beyond the vision, it is necessary to switch and calculate back and forth between the camera coordinate system, the chip coordinate system and the mechanical module, introducing a large error. However, in this solution, by infinitely expanding the virtual vision of the camera, this cumbersome and error-prone switching process is avoided. First, a dual coordinate system is defined and feature points are extracted, a homogeneous coordinate matrix is established to complete coordinate transformation and distortion correction, and then new feature points are continuously selected to achieve infinite expansion of the vision to cope with a large range of chip substrates. Then, the mounting position deviation is analyzed and the error nature is judged, and the error compensation is respectively carried out by re-calibrating the H matrix parameters or the method of taking the average of multiple measurements. Finally, the compensated coordinates are output to the moving module of the mounter to complete the mounting. This solution improves the mounting efficiency and accuracy and reduces the error.

[0018] 2. The method for compensating the error of the camera field of view of the infinitely expandable mounter accurately defines the chip substrate and camera coordinate systems in the establishment of the coordinate system and the selection of feature points. The SIFT algorithm is used to extract feature points with scale and rotation invariance, providing a reliable basis for subsequent coordinate transformation and ensuring the accurate matching of feature points from different perspectives. The establishment of the homogeneous coordinate matrix H clearly describes the transformation relationship between coordinate systems. Combined with polynomial distortion model correction, it effectively eliminates the influence of camera optical distortion on coordinate transformation and improves the accuracy of coordinate transformation. By continuously selecting new feature points, the infinite expansion of the camera field of view is realized, solving the problem that it is difficult for the small camera field of view to cover a large range of chip substrates. There is no need to switch coordinate systems back and forth, improving the processing efficiency. Error analysis can accurately locate the error source and judge its nature, and compensation is carried out by adopting strategies such as recalibrating the parameters of the H matrix, taking the average of multiple measurements, and data smoothing processing, reducing the mounting error. Finally, the compensated coordinates are output to the moving module of the mounter, realizing the high-precision mounting of the chip at the specified position on the substrate, improving the mounting quality and production efficiency, and reducing the defective rate.

[0019] 3. The method for compensating the error of the camera field of view of the infinitely expandable mounter can accurately judge the nature of the error based on error analysis. If there is a systematic error, its source can be identified, such as mechanical installation, camera calibration, or environmental factors, and then targeted measures can be taken to correct it. If it is a random error, the reasons for its generation can be understood, such as measurement noise, electromagnetic interference, etc., and methods such as taking the average of multiple measurements are used to reduce the influence. This accurate judgment and effective processing of the nature of the error significantly improve the mounting accuracy, reduce the mounting deviation caused by errors, ensure the chip mounting quality, reduce the defective rate, and improve the production efficiency and product reliability. Brief Description of the Drawings

[0020] Figure 1 is a schematic flowchart of the steps of the present invention; Figure 2 is a top view schematic diagram of the two coordinate systems of the present invention; Figure 3 is a three-dimensional schematic diagram of the two coordinate systems of the present invention. Detailed Embodiment

[0021] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0022] Refer to Figures 1 to 3 , the present invention provides the following three technical solutions: The first implementation method: A method for compensating errors by infinitely expanding the camera field of view of a chip mounter, including the following steps: S1. Define the chip substrate coordinate system and the camera coordinate system, automatically extract the feature points with scale and rotation invariance in the chip substrate image using the SIFT algorithm, and obtain the coordinates of the feature points in the two coordinate systems. S2. Establish a homogeneous coordinate matrix H to describe the transformation relationship from the chip substrate coordinate system to the camera coordinate system, transform the points in the chip substrate coordinate system to the camera coordinate system through the matrix H, and consider the camera optical distortion. Use a polynomial distortion model to correct the transformed coordinates to eliminate the distortion effect. S3. Continuously select new feature points in the chip substrate coordinate system, repeat the coordinate transformation process, and achieve infinite expansion of the camera field of view to process the chip substrate beyond the actual camera field of view. S4. Analyze the deviation between the actual position and the theoretical position of the chip after placement, and judge the nature of the error. For systematic errors, recalibrate the parameters of the H matrix; for random errors, use the method of taking the average of multiple measurements. Apply the error compensation strategy to the coordinate transformation process and correct the transformed coordinates. S5. Output the error-compensated camera coordinates to the chip mounter moving module for chip placement work.

[0023] Due to the small camera field of view in the conventional method, when dealing with the chip placement positions beyond the field of view, it is necessary to switch back and forth between the camera coordinate system, the chip coordinate system, and the mechanical module for calculation, introducing a large error. However, this solution avoids this cumbersome and error-prone switching process by infinitely expanding the virtual field of view of the camera. First, define the dual coordinate systems and extract the feature points, establish the homogeneous coordinate matrix to complete the coordinate transformation and distortion correction, then continuously select new feature points to achieve infinite expansion of the field of view to cope with a large range of chip substrates. Then analyze the placement position deviation and judge the nature of the error, and use the method of recalibrating the parameters of the H matrix or taking the average of multiple measurements for error compensation respectively. Finally, output the compensated coordinates to the chip mounter moving module to complete the placement. This solution improves the placement efficiency and accuracy and reduces the error.

[0024] The second implementation method, the main difference from the first implementation method is that: the S1 specifically includes coordinate system establishment and feature point selection: S1.1 Establish the coordinate system: Define the chip substrate coordinate system , the origin O b is selected at one of the corner points of the chip substrate, the X b axis and the Y b axis are respectively along two sides of the chip substrate; Define the camera coordinate system , the origin O c is the camera optical center, the Xc The X-axis and the Y-axis c are respectively parallel to the horizontal and vertical directions of the camera imaging plane; Feature point selection: Use the SIFT feature point detection algorithm to process the chip substrate image, and automatically extract feature points with scale invariance and rotation invariance; Suppose n feature points are detected from the chip substrate image, and their coordinates in the chip substrate coordinate system are represented as , and their corresponding coordinates in the camera coordinate system are represented as .

[0025] The homogeneous coordinate matrix H in S2 is: ; where λ is the scaling factor, representing the scale transformation ratio from the chip substrate coordinate system to the camera coordinate system; θ is the rotation angle, representing the rotation relationship between the two coordinate systems; and are respectively the translation amounts of the feature point in the X and Y directions of the two coordinate systems.

[0026] In terms of coordinate system establishment and feature point selection, the chip substrate and camera coordinate systems are accurately defined, and the SIFT algorithm is used to extract feature points with scale and rotation invariance, providing a reliable basis for subsequent coordinate transformation and ensuring accurate matching of feature points from different perspectives.

[0027] S3 specifically includes coordinate transformation and field of view expansion: S3.1 Coordinate transformation: For any point in the chip substrate coordinate system, its homogeneous coordinate representation is , and its coordinates transformed to the camera coordinate system through the homogeneous coordinate matrix H are: ; Considering the optical distortion of the camera, a polynomial distortion model is used for correction; suppose the distortion model is: ; ; where , k1, k2, k3... are distortion coefficients determined through calibration experiments; S3.2 Field of view expansion: Continuously select new feature points in the chip substrate coordinate system and repeat the above coordinate transformation process to achieve infinite expansion of the camera's field of view.

[0028] The establishment of the homogeneous coordinate matrix H clearly describes the transformation relationship between coordinate systems. Combining with the polynomial distortion model correction, it effectively eliminates the influence of camera optical distortion on coordinate transformation and improves the accuracy of coordinate transformation. By continuously selecting new feature points, the camera's field of view is infinitely expanded, solving the problem that it is difficult for the camera's small field of view to cover a large-range chip substrate. There is no need to switch coordinate systems back and forth, improving the processing efficiency.

[0029] The error analysis step in S4 includes: S4.1 Analyze the error source by actually measuring the deviation between the actual position and the theoretical position of the mounted chip; Let the theoretical position be , and the actually measured position be , then the error , ; S4.2 Analyze the nature of the error to determine whether it is a systematic error or a random error; Systematic errors are regular, such as those caused by mechanical installation errors, camera calibration errors, etc.; Random errors are random, such as measurement noise, etc.

[0030] The error compensation strategy in S4 includes: For systematic errors, if it is found that there is a fixed deviation related to the H matrix (for example, the translation amounts T x , T y have deviations), correct it by re-calibrating the parameters in the H matrix; Let the deviations in the translation amounts be ΔT x and ΔT y , and the included angle between the two coordinate systems be θ (θ is caused by the torsional error of the camera's own installation or the deflection transmitted from the chip substrate), then the corrected homogeneous coordinate matrix is: ; For random errors, use the method of taking the average of multiple measurements and data smoothing processing to reduce the influence; Error compensation implementation: Apply the formulated error compensation strategy to the coordinate transformation process to correct the transformed coordinates.

[0031] Error analysis can accurately locate the error source and judge the nature, and adopt strategies such as re-calibrating the H matrix parameters or taking the average of multiple measurements and data smoothing processing for compensation, reducing the mounting error.

[0032] S5 specifically includes: S5.1 Coordinate output: Output the coordinates (x c ′, y c ′) in the camera coordinate system after error compensation and distortion processing to the moving module of the mounter; S5.2 Moving and Mounting: According to the received coordinate information, the mounter controls the moving module to accurately mount the chip at the specified position on the substrate.

[0033] Finally, the compensated coordinates are output to the moving module of the mounter, achieving high-precision mounting of the chip at the specified position on the substrate, improving the mounting quality and production efficiency, and reducing the defective rate.

[0034] The third implementation method is mainly different from the first implementation method in that in the error property analysis, the error regularity analysis includes: S4.2.1 Data Collection: Multiple Measurements: Under the same experimental conditions, the same mounting position is measured m times, and the actual position coordinates of each measurement are recorded. ; S4.2.2 Analyzing Error Regularity: Calculating the Mean and Variance of Errors: Mean: ; ; Variance: ; ; Drawing an Error Distribution Diagram: Taking the measurement number i as the horizontal axis and the errors and as the vertical axis, draw a graph of the change of errors with the measurement number, or draw a histogram or scatter plot of the errors to observe the distribution of errors; Calculating the Autocorrelation Coefficient: The autocorrelation coefficient is used to reflect the correlation between errors at different time points (or measurement numbers): ; where represents the covariance of Δx i and , represents the standard deviation of , and a represents the lag order.

[0035] Systematic Error Judgment includes: If the error mean or is significantly non-zero, and the error distribution diagram shows an obvious trend or periodic change in the errors, or the autocorrelation coefficient is significantly non-zero at a certain lag order a, it indicates the existence of systematic errors; Systematic errors may be caused by mechanical installation errors, camera calibration errors, environmental factors (such as temperature and humidity changes), etc.; Random Error Judgment includes: If the error mean or is close to zero, and the error distribution diagram shows that the errors are randomly distributed without obvious trends or periodic variations. At the same time, the autocorrelation coefficient is close to zero for all lag orders k, indicating the presence of random errors; random errors may be caused by measurement noise, electromagnetic interference, operator differences, etc.

[0036] By measuring the same mounting position multiple times and recording the actual position coordinates, it provides a sufficient and reliable data basis for error analysis. Calculating the error mean and variance can intuitively understand the overall level and dispersion degree of the errors, and initially judge the error distribution characteristics. Drawing the error distribution diagram presents the changes or distributions of the errors with the number of measurements in an intuitive graphical way, facilitating the discovery of whether there are abnormalities such as trends and periodicity in the errors. Calculating the autocorrelation coefficient can further explore the correlation between errors at different time points (or measurement times) and judge whether the errors have internal associations.

[0037] Based on these analyses, the nature of the errors can be accurately judged. If there are systematic errors, their sources can be identified, such as mechanical installation, camera calibration, or environmental factors, etc., and then corresponding measures can be taken to correct them; if they are random errors, the causes can be understood, such as measurement noise, electromagnetic interference, etc., and methods such as taking the average of multiple measurements can be used to reduce the impact. This accurate judgment and effective handling of the nature of the errors significantly improve the mounting accuracy, reduce the mounting deviation caused by errors, ensure the chip mounting quality, reduce the defective rate, and improve the production efficiency and product reliability.

[0038] Example: After our factory adopted this technical solution, compared with the original solution, the comparison data is shown in Table 1: Table 1 Comparison Table of Data between the New and Old Solutions

[0039] As shown above: Mounting accuracy: The existing solution has significantly improved in mounting accuracy, from the original 15 ± 3μm to 5 ± 1μm.

[0040] Field of view: The field of view of the original solution was fixed at 400 mm² (20x20 mm), while the existing solution can theoretically be infinitely expanded.

[0041] Processing time: The processing time of the existing solution has been significantly reduced, from the original 60 seconds / substrate to 30 seconds / substrate.

[0042] Analysis of error sources: The existing solution classifies the errors into systematic errors and random errors and effectively compensates for them respectively, resulting in a significant reduction in both types of errors.

[0043] Error Compensation Strategy: Existing solutions have introduced compensation strategies for systematic and random errors, improving the accuracy of chip mounting.

[0044] System Complexity: By simplifying the process and reducing the number of coordinate system switches, the system complexity of existing solutions has been significantly reduced.

[0045] Defective Rate: The improvement in chip mounting accuracy has directly led to a significant reduction in the defective rate, from the original 5% to 1%.

[0046] Maximum Processable Substrate Size: Existing solutions can process larger-sized chip substrates and are theoretically not limited by the field of view.

[0047] Calibration Time and Accuracy: The calibration time of existing solutions has been reduced by half, and the calibration accuracy has also been improved.

[0048] Long-term Stability: Existing solutions have also seen a significant improvement in long-term stability, with the trouble-free operation time increasing from the original 100 hours to 500 hours.

[0049] Meanwhile, the content not described in detail in this specification belongs to the prior art well-known to those skilled in the art.

[0050] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variation thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article or device.

[0051] Although the embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for infinitely expanding the field of view of a camera of a chip mounter for error compensation, characterized in that: The following steps are involved: S1. Define the chip substrate coordinate system and the camera coordinate system, use the SIFT algorithm to automatically extract feature points with scale and rotation invariance in the chip substrate image, and obtain the coordinates of the feature points in the two coordinate systems; S2. Establish a homogeneous coordinate matrix H to describe the transformation relationship from the chip substrate coordinate system to the camera coordinate system. Use the matrix H to transform the points in the chip substrate coordinate system to the camera coordinate system, and consider the camera optical distortion. Use a polynomial distortion model to correct the transformed coordinates to eliminate the distortion effect. S3, continuously selecting new chip substrate coordinate system feature points, repeating the coordinate transformation process, and realizing infinite expansion of the camera field of view to process chip substrates beyond the actual field of view of the camera; S4. Analyze the deviation between the actual position and theoretical position of the chip after mounting to determine the nature of the error; recalibrate the H matrix parameters for systematic errors; use the average method of multiple measurements for random errors; apply the error compensation strategy to the coordinate transformation process to correct the transformed coordinates; S5. Output the error-compensated camera coordinates to the placement machine mobile module for chip placement.

2. The method for infinitely expanding the field of view of a chip mounter camera for error compensation according to claim 1, characterized in that: S1 specifically includes coordinate system establishment and feature point selection: S1.1 Establish the coordinate system: Define the chip-substrate coordinate system , origin O b Select one of the corner points of the chip substrate, X b Axis and Y b The axes are respectively along the two edges of the chip substrate; Define the camera coordinate system , origin O c is the camera optical center, X c Axis and Y c The axes are parallel to the horizontal and vertical directions of the camera imaging plane, respectively; Feature point selection: The chip substrate image is processed using the SIFT feature point detection algorithm to automatically extract feature points with scale invariance and rotation invariance; Assume that n feature points are detected from the chip substrate image, and their coordinates in the chip substrate coordinate system are expressed as , the corresponding coordinates in the camera coordinate system are expressed as .

3. The method for infinitely expanding the field of view of a chip mounter camera for error compensation according to claim 2, characterized in that: The homogeneous coordinate matrix H in S2 is: ； Among them, λ is the scaling factor, which indicates the scale transformation ratio from the chip substrate coordinate system to the camera coordinate system; θ is the rotation angle, which indicates the rotation relationship between the two coordinate systems; and They are the translation amounts of the feature points in the X and Y directions of the two coordinate systems respectively.

4. The method for infinitely expanding the field of view of a chip mounter camera for error compensation according to claim 3, characterized in that: The S3 specifically includes coordinate transformation and field of view expansion: S3.1 Coordinate transformation: For any point in the chip substrate coordinate system , the homogeneous coordinates are expressed as , transformed to the coordinates in the camera coordinate system through the homogeneous coordinate matrix H for: ； Considering the optical distortion of the camera, a polynomial distortion model is used for correction; let the distortion model be: ； ； in, , k1, k2, k3... are distortion coefficients, determined by calibration experiments; S3.2 Field of view expansion: continuously select feature points in the new chip substrate coordinate system and repeat the above coordinate transformation process to achieve infinite expansion of the camera field of view.

5. The method for infinitely expanding the field of view of a camera of a chip mounter for error compensation according to claim 1, characterized in that: The error analysis step in S4 includes: S4.1 Analyze the source of error by measuring the deviation between the chip position after mounting and the theoretical position; assuming the theoretical position is The actual measurement position is , then the error , ; S4.2 Analyze the nature of the error and determine whether it is a systematic error or a random error; systematic errors are regular, and random errors are random.

6. The method for infinitely expanding the field of view of a chip mounter camera for error compensation according to claim 5, characterized in that: The error compensation strategy in S4 includes: For systematic errors, if a fixed deviation related to the H matrix is ​​found, correction is made by recalibrating the parameters in the H matrix; let the deviation of the translation be ΔT x and ΔT y , then the modified homogeneous coordinate matrix is: ； For random errors, the impact is reduced by taking average of multiple measurements and smoothing the data. Error compensation implementation: Apply the formulated error compensation strategy to the coordinate transformation process and correct the transformed coordinates.

7. The method for infinitely expanding the field of view of a chip mounter camera for error compensation according to claim 1, characterized in that: The S5 specifically includes: S5.1 Coordinate output: The coordinates (x c ′,y c ') Output to the mobile module of the placement machine; S5.2 Mobile placement: The placement machine controls the mobile module to accurately place the chip at the specified position on the substrate based on the received coordinate information.

8. The method for infinitely expanding the field of view of a chip mounter camera for error compensation according to claim 5, characterized in that: In the error property analysis, the error regularity analysis includes: S4.2.1 Data collection: Multiple measurements: Under the same experimental conditions, measure the same mounting position m times and record the actual position coordinates of each measurement. ; S4.2.2 Analytical error regularity: Compute the mean and variance of the error: Mean: ； ； variance: ； ； Draw the error distribution graph: take the number of measurements i as the horizontal axis and the error and As the vertical axis, plot the error versus the number of measurements, or plot a histogram or scatter plot of the error to observe the distribution of the error; Calculate the autocorrelation coefficient: Autocorrelation coefficient Used to reflect the correlation between errors at different time points or measurement times: ； in, Denotes Δx i and The covariance of express The standard deviation of , a represents the lag order.

9. The method for infinitely expanding the field of view of a camera of a chip mounter for error compensation according to claim 8, characterized in that: Systematic error judgment includes: if the error mean or Significantly different from zero, and the error distribution graph shows that the error has an obvious trend or periodic change, or the autocorrelation coefficient If it is significantly different from zero at a certain lag order a, it indicates the existence of systematic errors.

10. The method for infinitely expanding the field of view of a camera of a chip mounter for error compensation according to claim 1, characterized in that: Random error judgment includes: if the error mean or is close to zero, and the error distribution graph shows that the errors are randomly distributed, with no obvious trend or periodic changes. If it is close to zero at all lag orders k, it indicates the presence of random errors.

Images