Height correction and monitoring method based on ceramic planar substrate

Abstract

The invention discloses a height correction and monitoring method and system based on a ceramic plane substrate and a storage medium. The core of the method is that firstly, a ceramic plane calibration plate is utilized, full-scale dense sampling is carried out by controlling Z-axis precise movement of equipment, and a high-precision phase height mapping model is established pixel by pixel so as to effectively compensate nonlinear errors of a system; then, an independent ceramic step verification block is adopted for instant verification, and the calibration quality is ensured through quantitative evaluation; and finally, the verification process is executed regularly and automatically, trend analysis is performed based on time sequence data, and long-term performance monitoring and early warning of the equipment are realized. The system comprises a control processing unit for executing the method and a special calibration object. According to the invention, automation and high precision of a calibration process are realized, and long-term reliability of a measurement result is ensured through continuous monitoring and intelligent diagnosis of an equipment state.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of precision measurement and instrument technology, and particularly relates to a height correction and monitoring method based on a ceramic plane substrate. BACKGROUND

[0002] Structured light three-dimensional imaging technology is one of the key technologies in the field of modern precision measurement. The basic principle is as follows: a coded light fringe pattern is projected onto the surface of a measured object, a camera captures the deformed fringe image due to the three-dimensional topography modulation of the object surface, then the absolute phase information is recovered through phase calculation technology, and finally the three-dimensional topography of the object is reconstructed according to an accurate "phase-height" mapping model. In this technology, the accuracy of the "phase-height" mapping model is the core of the measurement accuracy, and the model must be established through a precise calibration process.

[0003] At present, the conventional method for establishing the above-mentioned mapping model in the industry mainly adopts a method of using a reference block (such as a gauge block) with a known height or moving a plane calibration plate on a precision guide rail. However, these existing methods have obvious limitations, which restrict the further improvement of measurement accuracy and system reliability: The calibration reference is dependent on a single reference, and it is difficult to ensure the consistency of the full-range accuracy: Most traditional methods only use one or a limited number of discrete height points for calibration. The mapping model established by this sparse sampling method cannot accurately depict the possible nonlinear distortion of the optical system in the entire measurement range, resulting in a significant decrease in measurement accuracy at non-calibration points and inability to ensure the uniformity of accuracy in the full-range range.

[0004] The calibration process is complicated, and the degree of automation and efficiency is low: When using gauge blocks as calibration references, different specifications of block gauges need to be manually replaced and aligned repeatedly, which is low in operation efficiency and is prone to introduce human operation errors. Although the method of using a high-precision one-dimensional displacement platform to move the calibration plate can achieve automation, the platform itself is high in cost, which greatly increases the system cost and makes it difficult to be popularized and applied in industrial fields.

[0005] There is a lack of effective online verification and long-term performance monitoring mechanism: Traditional calibration is considered as a one-time, static process. During long-term use, due to environmental factors such as vibration, temperature fluctuation and aging of optical elements, the calibration parameters will drift, resulting in gradual degradation of measurement accuracy. However, the existing technology lacks a built-in and convenient mechanism to verify the calibration effect in real time, and it is even more difficult to continuously monitor and warn the long-term performance of the device. This makes it difficult to find the accuracy decay in time, seriously affecting the long-term reliability and traceability of the measurement results, and even may lead to batch product quality problems.

[0006] In summary, existing calibration techniques are insufficient to meet the comprehensive requirements of high-precision 3D measurement for full-range accuracy, automation efficiency, and long-term reliability. Therefore, there is an urgent need in this field for an integrated solution capable of achieving high-precision automated calibration across the entire measurement range, with both immediate verification and long-term performance monitoring capabilities, to overcome the shortcomings of the existing technologies. Summary of the Invention

[0007] The purpose of this invention is to provide a height correction and monitoring method based on a ceramic planar substrate. This invention achieves automation and high precision in the calibration process, and ensures the long-term reliability of the measurement results through continuous monitoring and intelligent diagnosis of the equipment status.

[0008] The technical solution adopted in this invention is as follows: A height correction and monitoring method based on a ceramic planar substrate, applied to a structured light 3D imaging device, includes the following steps: Full-range calibration procedure: Place the ceramic plane calibration plate in the equipment's field of view, and control the equipment's Z-axis from the initial position. Start with a fixed step size Gradually move to the maximum measurement range In each position ( Acquire deformed stripe images ; Calculate the absolute phase value pixel by pixel for the image sequence. For each pixel Using its calibration dataset An independent height mapping model is established through least squares fitting. ;in, For high-precision reference zero point, To calibrate the sampling step size, This is the maximum measuring height of the equipment. For the first The known physical height of each sampling point In order to be at a high altitude The grayscale image captured at that location. To extract from the image The absolute phase value obtained from the calculation; Instant verification and long-term monitoring steps: Replace the ceramic planar calibration plate with one of known height. The ceramic step verification block was used to obtain the test phase diagram through measurement. The height mapping model is used to calculate the three-dimensional point cloud, and the average measured height of each step is obtained through plane fitting. ; Calculate the absolute error Perform immediate verification; perform this periodically, based on time series data. Conduct trend analysis to achieve long-term monitoring and early warning; among which, For the first step verification block The known true height of each step, To verify the phase map obtained during the measurement, For the first The average measured height is obtained by performing planar fitting on the point cloud of each step. For the first The absolute error in the measurement of each step, In order to cut off the time For the first The height value was obtained by monitoring and measuring each step.

[0009] Preferably, in the full-range calibration step, the absolute phase is calculated. Methods include phase-shifting; when using the four-step phase-shifting method, it is based on four acquired images. , , , It can be calculated using the following formula: ;in, , , , These are four-step phase shift methods at the same height position. Four phase-shifted images were acquired.

[0010] Preferably, the height mapping model Using the reference plane phase The cubic polynomial model based on this is expressed as: ;in, For the physical height to be determined, The observed absolute phase, To be at a high reference zero point The reference phase value obtained from the measurement calculation is as follows: , , , These are the polynomial coefficients for this pixel, determined through fitting.

[0011] Preferably, the polynomial coefficients are obtained by least squares fitting; for each pixel, its coefficient vector... By minimizing its calibration dataset Sum of squared residuals To determine.

[0012] Preferably, the reference plane phase The methods for determining include: The phase value obtained from the acquisition and calculation is set as follows: .

[0013] Preferably, the trend analysis in the long-term monitoring step includes: monitoring data sequences for each step. Perform linear regression and calculate its slope. ;like If the absolute value of the value is statistically significantly greater than zero, then a precision drift trend is determined to exist; among them, For the first The regression slope of the height measurement of each step over time is used to characterize the trend and rate of its drift.

[0014] Preferably, by comparing the regression slopes of different steps Perform fault diagnosis: If different steps Significant differences in values ​​indicate nonlinear distortion in the optical system.

[0015] Preferably, the initial position The determination method includes: evaluating image sharpness through an evaluation function, positioning the surface of the ceramic plane calibration plate at the optimal focal plane position along the Z-axis, and setting this position as the height reference zero point. .

[0016] A non-volatile computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method.

[0017] A structured light 3D imaging device includes a projector, a camera, a Z-axis movement mechanism, and a control processing unit. The control processing unit is configured to execute the method described above and stores information for each pixel. The polynomial coefficient vector obtained by fitting and reference phase .

[0018] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are: This invention constructs a closed-loop technology system integrating "full-range calibration, real-time verification, and long-term monitoring," and combines it with dedicated storage media and equipment configuration to produce synergistic and beneficial effects, mainly reflected in the following aspects: 1. This invention uses a high-precision ceramic planar substrate as a reference. It performs dense sampling across the entire measurement range of the device and establishes an independent height mapping model (such as a cubic polynomial) for each pixel. This method can accurately characterize and effectively compensate for spatial nonlinear errors caused by lens distortion, viewing angle differences, and system geometry, thereby overcoming the shortcomings of traditional single-point or sparse-point calibration methods that experience a sharp drop in accuracy at non-calibrated points. As a result, it ensures that the measurement system maintains uniformly high measurement accuracy from the near field to the far field, significantly improving the overall accuracy and reliability of the 3D reconstruction data.

[0019] 2. The entire calibration, verification, and monitoring process is completed automatically by the control processing unit, eliminating the need for manual intervention to replace gauge blocks or perform complex calculations. This highly automated design not only frees operators from tedious and specialized tasks, avoiding human error, but also makes factory calibration and periodic on-site maintenance of batch equipment efficient and fast. This is particularly suitable for large-scale industrial applications, greatly improving production and maintenance efficiency.

[0020] 3. The system's built-in real-time verification mechanism provides objective quality judgments for each calibration result, ensuring the reliability of the measurement starting point. More importantly, through long-term monitoring and trend analysis functions, the system can promptly detect systematic drift trends in accuracy, achieving a paradigm shift from "passive maintenance" to "proactive prevention." This effectively avoids batch quality incidents caused by the hidden degradation of equipment performance, guarantees the long-term reliability and traceability of measurement data throughout its entire lifecycle, and meets the stringent quality control requirements of high-end manufacturing industries.

[0021] 4. The time-series data accumulated through long-term monitoring provides a data foundation for equipment condition assessment and fault pre-diagnosis. By analyzing the differences in drift characteristics at different elevation points, the system can provide key clues for potential fault root causes (such as systematic errors or nonlinear distortions). This function elevates the measurement equipment from a traditional "black box" tool to an intelligent system with preliminary self-diagnostic capabilities, significantly reducing the difficulty and time cost of fault diagnosis and optimizing the full lifecycle operation and maintenance management of the equipment.

[0022] In summary, through the synergistic innovation of methods, media, and equipment, this invention has successfully constructed a high-precision, highly reliable, self-diagnostic, and easy-to-use three-dimensional measurement solution, effectively solving many pain points in existing technologies and demonstrating significant technological advancements and broad application prospects in the field of precision measurement. Attached Figure Description

[0023] Figure 1 This is a schematic diagram of the process of the present invention; Figure 2 This is a flowchart illustrating the full-range calibration steps of the present invention. Figure 3 This is a flowchart illustrating the real-time verification and long-term monitoring steps of the present invention. Figure 4 This is a schematic diagram of the installation of the system of the present invention. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0025] See Figures 1-4 This invention relates to a height correction and monitoring method based on a ceramic planar substrate, which is applied to a structured light 3D imaging device. The method includes a full-range calibration step and an instant verification and long-term monitoring step.

[0026] The full-range calibration step is the core foundation and primary step for achieving high-precision measurement in this method. Its fundamental purpose is to establish a precise, pixel-level mathematical model for each specific structured light 3D imaging device, reliably converting the phase information calculated from the camera-acquired images into true physical height information. The quality of this step directly determines the final measurement accuracy and reliability of the entire system.

[0027] Any precise measurement requires a stable and accurate reference datum. This step begins by placing the ceramic calibration plate at the center of the device's measurement field of view. The choice of ceramic material is crucial due to its extremely high dimensional stability, extremely low coefficient of thermal expansion, and excellent wear resistance, providing a long-term reliable physical reference plane with extremely high flatness (e.g., better than 1 micrometer).

[0028] Before starting systematic data acquisition, the measurement "origin," i.e., the altitude reference zero point, must be determined. This is achieved through the principle of autofocus. The control system slowly moves the device's Z-axis while the camera continuously captures images of the calibration plate surface. The system calculates the sharpness evaluation function value (e.g., the Tenengrad gradient function) for each image in real time, which reaches its maximum when the image is sharpest. The control system searches for the Z-axis position that maximizes the sharpness evaluation value and precisely sets this position as the zero point of the height reference. This operation not only defines the spatial geometric origin, but its more crucial significance lies in the fact that the phase map acquired and calculated at this location will be defined as the reference phase field for each pixel. All subsequent calculations will be based on this reference phase, thus effectively eliminating background interference.

[0029] After establishing the baseline, the intensive data acquisition process, which truly embodies the concept of "full range," began. The control system will drive the Z-axis from... Begin with a preset, sufficiently small fixed step size. (e.g., 0.5 mm), gradually and precisely move to the entire measuring range of the device. (For example, 15 mm). This results in a series of discrete sampling points with known locations along the entire height direction: ,in ( This allows for dozens of sampling points (N), ensuring adequate data representation throughout the entire measurement range. At each specific height location... The system executes a complete image acquisition process: the projector projects a set of preset coded structured light patterns (e.g., four sinusoidal fringe patterns with phase shifts of 90 degrees each, required by the four-step phase-shifting method) onto the ceramic calibration plate, and the camera is simultaneously triggered to acquire the corresponding deformed fringe images. For each The present invention ultimately yields an image sequence, such as four images. After this process is completed, the present invention obtains a set of image sequences covering the entire Z-axis range. Each set of images corresponds to a precisely known physical height. .

[0030] The acquired raw image is grayscale information and needs to be converted into phase information directly related to height. This will be illustrated using the four-step phase-shifting method described above. For images with height... A set of four images collected at [location] arrive The absolute phase value is calculated pixel by pixel using the following formula. : ; By using difference and ratio operations, the effects of uneven background light intensity and differences in surface reflectivity in the image are effectively eliminated. The calculated initial phase is wrapped in... The entangled phases within the interval need to be restored to continuous and unique absolute phase values ​​using a phase unrolling algorithm. Thus, the present invention per height Each obtained a corresponding absolute phase field distributed across the entire field. IV. Pixel-level Mapping Model Establishment: Implementation of the Core Algorithm This is the most crucial step in full-range calibration, aiming to establish a pixel-level mapping model for each pixel. Each pixel independently establishes its own phase-height mapping model. Following the aforementioned steps, for any pixel in the field of view, this invention obtains a unique calibration dataset: This set of data constitutes the "learning sample" for that pixel.

[0031] Next, a cubic polynomial model is used to fit this mapping relationship. The specific form of the model is as follows: ; in: It is the physical height to be determined. It is the observed absolute phase value. Is this pixel at the reference zero point? The reference phase value calculated at that location. These are the polynomial coefficients for the pixel to be determined. A cubic polynomial was chosen because it strikes a good balance between computational complexity and fitting capability, and is sufficient to accurately describe the nonlinear distortion present in most optical systems. The purpose of using variables is to improve the stability of numerical calculations.

[0032] The model parameters are determined through least squares fitting. For each pixel, a set of coefficient vectors is found. This makes the predicted value of the polynomial on all calibration data points... Compared with the actual height value Minimize the sum of squared residuals between: By solving this optimization problem, the optimal fitting coefficient for that pixel can be obtained.

[0033] The final product of full-range calibration is not a simple conversion table, but a massive parameter database. For the millions or even tens of millions of pixels in the device's field of view, each point stores its own two types of parameters: reference phase value; And the polynomial coefficient vector: This set of parameters constitutes the device's unique "digital DNA" or "fingerprint." When actually measuring any unknown object, the system only needs to first calculate the phase diagram of the object being measured. Then for each pixel in the image By substituting the measured phase value into the pre-stored polynomial model corresponding to that point, the true physical height of that point can be calculated in real time and in parallel. This allows for the rapid reconstruction of the entire object's three-dimensional shape.

[0034] The full-range calibration step, through systematic intensive sampling and pixel-level precise modeling, fundamentally overcomes the limitations of traditional calibration methods (such as single-point or sparse-point calibration). It accurately compensates for the nonlinear errors of the optical system, ensuring uniform and consistent high accuracy across the entire measurement range. This not only provides reliable calibration results for subsequent on-the-fly verification but also lays a traceable data foundation for long-term performance monitoring, serving as the cornerstone for achieving the method's automated, highly reliable, and intelligent diagnostic capabilities.

[0035] See Figure 3 The immediate verification step is performed immediately after the full-range calibration is completed. Its core purpose is to determine whether the calibration was truly successful and whether the current measurement accuracy of the equipment meets the expected standards. This is a final check on the calibration results and the last quality check before the equipment is put into formal use.

[0036] The authority of the verification comes from a physical benchmark with a higher level of accuracy, independent of the calibration process. The scheme uses a ceramic stepped verification block. Like the ceramic planar calibration plate, it possesses excellent thermal and mechanical stability, but its structure is more complex, typically containing multiple (e.g., three) blocks with known and precisely traceable true height values. The planar step (where j is the step index) can be denoted as . For example, the step heights could be 1.000 mm, 2.000 mm, or 5.000 mm, with machining accuracy reaching the micrometer or even sub-micrometer level. The advantage of using step blocks instead of a single planar plate is that it allows for simultaneous inspection of multiple points at different heights, thereby comprehensively evaluating the accuracy consistency and linearity of the calibrated equipment across the entire measurement range.

[0037] The verification process begins by removing the calibration planar plate and replacing it with a stepped verification block. Subsequently, the system performs a standard 3D measurement procedure: a projector projects a structured light pattern, a camera acquires images, and a complete test phase map is calculated. Next, the system calls the height mapping model established for each pixel during the full-range calibration. Specifically, for each pixel in the phase map, its phase value is... Substituting the values ​​into the polynomial model corresponding to that point, its three-dimensional coordinates are calculated in real time, ultimately generating a representative value. The system verifies the 3D point cloud data of the step surface. After obtaining the point cloud, the system enters the crucial data processing stage. Since the ideal surface of each step is a plane, the system uses a least-squares plane fitting algorithm to fit an optimal plane equation from the cluttered point cloud corresponding to each step. The spatial position of this fitted plane accurately reflects the average height of the step, which is recorded as the measured height value in this invention. This step effectively eliminates the effects of random noise and surface micro-irregularities, resulting in a stable and reliable measurement result.

[0038] With the measured values, an authoritative comparison can be made with the known true values. For each step, its absolute error is calculated: ; This error value This is a direct quantitative representation of the measurement accuracy of the equipment in its current state. Next, the system will compare the error values ​​of all steps with a preset acceptable threshold (usually set based on equipment specifications or the accuracy requirements of the application scenario). If the absolute error of all steps... If all values ​​are below the acceptable threshold, the system determines that the calibration was successful, the equipment's accuracy is good, and it can be put into formal use. This provides users with strong confidence. If the error of any step exceeds the threshold, the system will immediately issue an alarm. This indicates that there may be a problem with the recently completed calibration (such as operational error, environmental interference), or a sudden equipment failure. In this case, the equipment should be taken out of service, and the cause needs to be investigated. Usually, a full-range calibration needs to be performed again.

[0039] The significance of instant verification lies in its use of an independent and objective "judge" to provide a quantitative and decisive pass / fail judgment on the instantaneous accuracy status of the equipment, ensuring that the equipment's "ruler" is accurate before any important measurement begins.

[0040] If real-time verification is like taking a "snapshot" of a device's accuracy, then long-term monitoring is like stringing these snapshots together into a "documentary," analyzing its development to predict future outcomes. Its core objective is to determine how the device's accuracy is changing over time, its future stability, and when maintenance is needed.

[0041] The long-term monitoring is achieved by periodically and automatically repeating the real-time verification process. This cycle can be flexibly set according to factors such as equipment usage intensity and environmental stability, for example, automatically once every day when the machine is turned on, at a fixed time every week, or after each certain production batch is completed.

[0042] After each monitoring and verification operation, the system not only determines whether the current operation is successful, but more importantly, it records the measured height value of each step j. Along with the current timestamp t, it is securely stored in the database as a data point. Thus, for each step, a time series of data will be generated over time. This data sequence has become a valuable "archive" of the evolution of equipment precision.

[0043] By performing trend analysis on these time series data, we can move from "perceiving the present" to "predicting the future." The core analytical method is linear regression. A linear fit is performed on the historical data series for each step j to obtain a straight line that best represents its trend. The slope of this line... It is a key indicator for trend analysis. The average rate of change of the measured height of the step over time was quantified.

[0044] This indicates that the measurement at this point is very stable and there is no systematic drift.

[0045] This indicates that the measured value has a gradually increasing trend of drift.

[0046] This indicates that the measured value has a gradually decreasing trend of drift.

[0047] The system will determine the observed slope through statistical tests (such as calculating the p-value). Whether it is significantly different from zero is used to confirm that this trend is a real system drift, rather than random fluctuation.

[0048] Based on trend analysis, long-term monitoring systems can achieve intelligent early warning: Early warning: The system can set a trend slope threshold. Once the drift rate of a certain step is detected... | Within the acceptable range, the system will also issue an early warning. This is equivalent to issuing a "slippery road ahead, please slow down" warning before the equipment's accuracy "falls off the cliff".

[0049] Predictive maintenance: based on the current drift rate By allowing margins for distance accuracy limits, the system can roughly estimate how much time (or usage cycle) the equipment has left before exceeding tolerance. This enables maintenance personnel to schedule maintenance (such as recalibration) during production breaks, thereby achieving predictive maintenance and minimizing unplanned downtime and product quality incidents caused by sudden equipment inaccuracies.

[0050] Even more powerfully, by comparing the drift trends of different steps, the system can also provide clues for fault diagnosis: Systematic drift: If all steps Similar values ​​and consistent direction usually indicate a global systematic error, such as uniform scaling of the entire measurement system due to changes in ambient temperature.

[0051] Nonlinear distortion: If steps of different heights... Significant differences in values ​​(e.g., high steps drift quickly, low steps drift slowly) strongly suggest nonlinear distortion in the optical system. Possible causes include non-uniform deformation of the lens due to stress or temperature gradients, or minute changes in the relative geometry of the camera / projector. This diagnostic capability provides valuable direction for quickly locating and resolving faults.

[0052] This system ensures that measurement accuracy is traceable, verifiable, predictable, and controllable throughout the entire lifecycle of the equipment. It transforms high-precision measurement equipment from a "black box" tool that relies on experience-based judgment into a "transparent" intelligent partner capable of self-sensing, self-diagnosing, and self-predicting, providing crucial reliability and predictability guarantees for intelligent industrial manufacturing.

[0053] A non-volatile computer-readable storage medium stores a computer program thereon, which, when executed by a processor, implements the steps of the method. This medium refers to a physical entity capable of persistently storing data even when power is off, constituting a fixed carrier for the computer program. Specific embodiments include, but are not limited to, read-only memory and flash memory in embedded systems, optical discs and solid-state drives used for software distribution, or any other storage device that meets the "non-volatile" requirement. The purpose of this medium is to ensure that the instruction set implementing the method of this invention can be stably and permanently stored and repeatedly accessed by computing devices.

[0054] A structured light 3D imaging device includes a projector, a camera, a Z-axis movement mechanism, and a control and processing unit. Specifically, the projection module projects an coded structured light pattern onto the object under test. The image acquisition module (camera) captures images of deformable stripes modulated on the object's surface. The Z-axis movement mechanism provides precise axial displacement, offering the necessary mechanical motion capability for full-range calibration.

[0055] The control processing unit is programmed or embedded to execute all the steps of the aforementioned "height correction and monitoring method based on ceramic planar substrate". This means that the unit embeds complete algorithmic logic and control flow to achieve full-range calibration, real-time verification, and long-term monitoring. For example, it includes instructions for controlling the Z-axis mechanism to move at preset step sizes, a calculation module for performing phase extraction using the phase-shifting method, an algorithm for least-squares fitting to establish a pixel-level height mapping model, and routines for plane fitting, error analysis, and trend warning. More importantly, it stores information for each pixel. The polynomial coefficient vector obtained by fitting and reference phase This parameter set, a "digital fingerprint" obtained by calibrating the device's optical system using the aforementioned method before the device leaves the factory, is a prerequisite for high-precision measurement. During the measurement process, the control processing unit drives the projection module and image acquisition module to acquire the phase information of the object being measured. Subsequently, instead of using a general algorithm, it calls its pre-stored pixel-level calibration parameters, precisely matched to the characteristics of this device, to perform a high-precision transformation on the observation value of each pixel. For the monitoring function, this unit automatically schedules the verification process and performs trend analysis based on historical data. The entire workflow demonstrates a deep integration of hardware execution with built-in dedicated software and dedicated data. The selection of the polynomial order is based on: First-order linear: Applicable to ideal optical systems, but with low computational cost. Third-order polynomials: Compensate for lens distortion and projection nonlinearity with moderate computational cost. Fifth-order polynomials: extremely nonlinear systems, but prone to overfitting and with relatively high computational cost. During the fitting process, a pixel-by-pixel fitting method is used to ensure that each pixel has high fitting accuracy.

[0056] To objectively evaluate the advantages of the "height correction and monitoring method based on ceramic planar substrate" proposed in this invention compared to traditional calibration methods, this comparative test was conducted. The test aims to verify the performance improvements of this invention in terms of calibration accuracy, efficiency, and long-term stability monitoring capabilities.

[0057] Specifically, the 3D imaging equipment is configured as follows: Industrial camera: 4096×3000s resolution; Projector: TI DLP4710, resolution 1920×1080; Lens: Telecentric lens, magnification 0.46×, distortion <0.05%, working distance 300mm; Ceramic calibration plate specifications: Material: 99.6% alumina ceramic, coefficient of thermal expansion 6.5×10 -6 / ℃; Planar calibration plate: dimensions 70mm×70mm×1mm, flatness <1μm, surface roughness Ra<0.1μm; Step verification block: Three-step design with heights of 1.000mm, 2.000mm, and 3.000mm respectively, tolerance ±0.5μm; Installation requirements: Level adjustable platform, flatness <5μm, fixed with rigid clamps; Motion control system: Z-axis platform: travel 30mm, resolution 1μm, repeatability ±1μm; Movement parameters: calibrated step distance 0.5mm, movement speed 5mm / s, acceleration 0.5m / s²; Experimental environment requirements: Temperature control: 20°C±1°C, temperature gradient <0.5°C / h; Vibration control: vibration isolation frequency <2Hz, vibration acceleration <0.01g; Humidity control: 45%±10% RH, anti-condensation; Lighting control: Dark room environment, ambient light <5 lux.

[0058] The table below details the specific data for the two methods under different test items:

[0059] The following is an example of monitoring data from this invention (10-day trend): Equipment 0.5mm step height drift trend

[0060] This invention achieves full automation. Although the calibration time for a single operation is slightly longer than the simplest single-point calibration, a single operation can complete high-precision calibration across the entire range, making it far more efficient than the traditional multi-point calibration method that requires multiple manual operations. In terms of accuracy, this invention demonstrates consistent high precision across the entire range. Traditional single-point methods exhibit significant errors at non-calibration points; while traditional multi-point methods offer improved accuracy, their precision and stability are still inferior to the continuous fitting model of this invention due to limitations in the number and distribution of calibration points.

[0061] The greatest advantage of this invention lies in its "long-term monitoring." Test simulations show that traditional methods can only detect accuracy deviations after a problem occurs (48 hours later), which is a "post-event remedy." In contrast, this invention, through continuous monitoring, detects minute systematic drift trends (+0.07μm / hour) in advance, achieving "early warning" and providing data support for predictive maintenance.

[0062] This invention effectively compensates for the nonlinear distortion of the optical system through dense sampling and high-order polynomial fitting, ensuring that the measurement error remains at a very low and uniform level at different heights. This is something that traditional methods cannot achieve.

[0063] Based on the above test data and analysis, the following conclusions can be drawn: The "Height Correction and Monitoring Method Based on Ceramic Planar Substrate" proposed in this invention significantly outperforms traditional calibration methods in terms of measurement accuracy (especially full-range consistency), long-term reliability, and automation. It not only provides a more precise calibration method but, more importantly, constructs a complete closed-loop quality control system capable of self-verification, trend prediction, and intelligent early warning. This method is particularly suitable for industrial testing and precision manufacturing scenarios with stringent requirements for measurement accuracy and long-term stability, effectively reducing quality risks and maintenance costs caused by equipment accuracy degradation.

[0064] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for height correction and monitoring based on ceramic planar substrate, applied to a structured light three-dimensional imaging device, characterized in that, The method comprises the steps of: Full range calibration procedure: place a ceramic flat calibration plate in the field of view of the device, control the device Z axis to move from the initial position Start with a fixed step size Move to the maximum measurement range step by step At each position ( )collect deformed fringe images ; Calculate the absolute phase value pixel by pixel for the image sequence ; For each pixel point , use its calibration data set to establish an independent height mapping model by least squares fitting ; Wherein, is the height reference zero point, is the calibration sampling step size, is the maximum measurement height of the device, is the known physical height of the th sampling point, is the gray image collected at height , is the absolute phase value calculated from image ; Instant verification and long-term monitoring: replace the ceramic planar calibration plate with ceramic step verification blocks with known heights , measure to obtain test phase maps ; calculate 3D point clouds using the height mapping model, and obtain the average measurement height of each step by plane fitting ; calculate the absolute error Instant verification; this is done periodically, based on time series data Trend analysis, long-term monitoring and early warning; wherein, is the known true height of the first step on the step verification block, is the phase map obtained in the verification measurement, is the average measurement height obtained by plane fitting on the first step point cloud, is the measurement absolute error of the first step, is the height value obtained by monitoring measurement on the first step at the time cut .

2. The ceramic planar substrate-based orthopedic alignment and monitoring method of claim 1, wherein, The full-range calibration step, in which the absolute phase is calculated The method includes a phase shift method; when a four-step phase shift method is used, the four images acquired , , , are calculated by the following formula: ;in, , , , These are four-step phase shift methods at the same height position. Four phase-shifted images were acquired.

3. The ceramic planar substrate-based orthopedic alignment and monitoring method of claim 1, wherein, The height mapping model employing a reference plane phase a cubic polynomial model referenced to the phase of the reference plane, expressed as: ; wherein is the physical height to be determined, is the observed absolute phase, is the reference phase value calculated at the height reference zero point , , , , is the polynomial coefficient of the pixel point determined by fitting.

4. The ceramic planar substrate-based orthopedic alignment and monitoring method of claim 3, wherein, The polynomial coefficients are solved by least squares fitting; for each pixel point, its coefficient vector is determined by minimizing the sum of squared residuals over its calibration dataset .​ 5. The ceramic planar substrate-based orthopedic alignment and monitoring method of claim 3, wherein, The reference plane phase The determination method comprises: The calculated phase value at the position is set as .

6. The ceramic planar substrate-based orthopedic alignment and monitoring method of claim 1, wherein, The trend analysis in the long-term monitoring step includes: monitoring data sequences for each stage. Perform linear regression and calculate its slope. ;like If the absolute value of the value is statistically significantly greater than zero, then a precision drift trend is determined to exist; among them, For the first The regression slope of the height measurement of each step over time is used to characterize the trend and rate of its drift.

7. The ceramic planar substrate-based orthopedic alignment and monitoring method of claim 6, wherein, By comparing the regression slopes of different steps Perform fault diagnosis: If different steps Significant differences in values ​​indicate nonlinear distortion in the optical system.

8. The ceramic planar substrate-based orthopedic alignment and monitoring method of claim 1, wherein, The initial position The determination method comprises: positioning the surface of the ceramic plane calibration plate at the best focal plane position of the Z axis by evaluating the image definition through an evaluation function, and setting the position as the height reference zero point .

9. A non-transitory computer readable storage medium having stored thereon a computer program, characterized in that, The computer program, when executed by a processor, implements the steps of the method of any one of claims 1 to 8.

10. A structured light three-dimensional imaging apparatus comprising a projector, a camera, a Z-axis moving mechanism, and a control processing unit, characterized by, The control processing unit is configured to perform the method of any one of claims 1 to 8 and stores for each pixel point the polynomial coefficient vector obtained from the fitting and the reference phase .

Images