Three-dimensional calibration plate for stereomicroscopic imaging system and performance evaluation method

Abstract

The present application relates to the field of microscopic imaging, and discloses a three-dimensional calibration plate for a stereoscopic microscopic imaging system and a performance evaluation method. The three-dimensional calibration plate is provided with at least two functional areas; each functional area has a specific three-dimensional geometric shape and surface texture; all the functional areas are defined by the same CAD model; and each functional area is used for collecting specific imaging evaluation indexes. The three-dimensional calibration plate provided by the present application can improve the calibration capability of the microscopic imaging system.

Description

Technical Field

[0001] This invention relates to the field of microscopic imaging, and more particularly to a three-dimensional calibration plate and performance evaluation method for stereoscopic microscopic imaging systems. Background Technology

[0002] Microscopic stereo imaging systems are a technique for acquiring the three-dimensional morphology of object surfaces at the millimeter to centimeter scale, and are widely used in precision manufacturing, microelectronics inspection, and life sciences. However, microscopic imaging itself has characteristics such as small depth of field, high magnification, strong surface reflection, sparse texture, and significant geometric distortion. These characteristics bring unique challenges to 3D reconstruction that are not present in ordinary vision systems. Simply relying on traditional camera calibration to obtain intrinsic and extrinsic parameters often fails to fully and accurately reflect the actual performance of the system in reconstructing complex 3D surfaces, especially in key indicators such as depth accuracy, morphological resolution, and spatial consistency.

[0003] Currently, commonly used performance evaluation methods mainly include: checkerboard-based planar calibration, reconstruction using non-standard physical objects (such as electronic components or biological samples), qualitative comparison relying on public datasets, and verification using self-made 3D calibration boards. These methods all have significant limitations. Checkerboard calibration primarily optimizes camera model parameters and reprojection errors, making it difficult to assess the reconstruction quality of complex 3D shapes; while non-standard physical objects closely resemble real-world applications, they lack known, traceable geometric truth values, resulting in poor repeatability and difficulty in quantitative comparison; dataset-based evaluations often rely on subjective visual comparisons, making it difficult to establish unified and objective quantitative standards; and self-made 3D targets often struggle to balance processing accuracy and surface texture, with evaluation metrics not closely related to target pattern design, and weak comparability across different systems.

[0004] Therefore, there is an urgent need to find a three-dimensional calibration plate and performance evaluation method for stereoscopic microscopy systems in order to improve the calibration capability of microscopy systems. Summary of the Invention

[0005] To address the technical problems in the prior art, this invention provides a three-dimensional calibration plate and performance evaluation method for stereoscopic microscopy imaging systems.

[0006] A three-dimensional calibration plate for a stereomicroscopic imaging system, the three-dimensional calibration plate having at least two functional areas;

[0007] Each of the aforementioned functional areas has a specific three-dimensional geometry and surface texture;

[0008] All the aforementioned functional areas are defined using the same CAD model;

[0009] Each of the aforementioned functional areas is used to collect specific imaging evaluation metrics.

[0010] Optionally, the functional area can be any one of the following: AFL planar area, periodic height modulation pattern area, stepped height area, two-dimensional pattern area, shape fidelity area, and auxiliary visual texture area.

[0011] Optionally, the AFL planar region is used to acquire imaging evaluation indicators including flatness deviation;

[0012] The periodic highly modulated pattern region is used to acquire imaging evaluation metrics including topographic spatial resolution.

[0013] The imaging evaluation indicators used for the stepped height region include depth magnification factor and linear deviation.

[0014] The two-dimensional pattern region is used to collect imaging evaluation indicators, including xy mapping deviation and lateral positioning error.

[0015] The morphology fidelity region is used to collect imaging evaluation indicators, including morphology fidelity.

[0016] The auxiliary visual texture region is used to collect imaging evaluation metrics including stereo matching stability and local feature visibility.

[0017] Optionally, the three-dimensional geometry of the AFL planar region includes a plane;

[0018] The three-dimensional geometry of the periodic highly modulated pattern region includes planar, radial star, multi-sector, multi-period groove, or stepped grating.

[0019] The three-dimensional geometry of the stepped height region includes a step structure that varies along the height direction;

[0020] The three-dimensional geometry of the two-dimensional pattern area includes 6×6 cross or grid markings;

[0021] The three-dimensional geometry of the morphological fidelity region includes sinusoidal surfaces, cylinders / semicircles, and / or triangular ridges;

[0022] The three-dimensional geometry of the auxiliary visual texture region includes random dots and diagonal stripes.

[0023] Optionally, the surface texture includes micron-scale surface microstructures; the surface microstructures are one or more of the following: triangular grooves, pyramidal structures, cones, pyramids, frustums, hexagonal pits, parallel grooves, and vertical grooves; and / or,

[0024] The surface texture is formed by laser processing.

[0025] A performance evaluation method for a stereomicroscopic imaging system includes:

[0026] Image data was acquired using the stereomicroscopic imaging system to be evaluated for any of the three-dimensional calibration plates mentioned above, and image measurement data was obtained.

[0027] The image measurement data is registered with the reference data to obtain registration measurement data;

[0028] The topographic data of each functional region is extracted from the registration measurement data, and the imaging evaluation index corresponding to the functional region is calculated based on the topographic data.

[0029] Optionally, the reference data includes a CAD model or a reference measurement model.

[0030] Optionally, the imaging evaluation metrics include topographic spatial resolution, depth magnification factor, linearity deviation, xy mapping deviation, lateral positioning error, topographic fidelity, stereo matching stability, and local feature visibility.

[0031] Optionally, after extracting the topographic data of each functional region from the registration measurement data and calculating the imaging evaluation index corresponding to the functional region based on the topographic data, the method further includes:

[0032] A comprehensive performance report is generated by statistically analyzing imaging evaluation indicators from multiple dimensions and under multiple conditions.

[0033] A method for manufacturing any of the three-dimensional calibration plates described above includes:

[0034] Based on the parameters and requirements of the stereomicroscopic imaging system to be evaluated, design the CAD model of the three-dimensional calibration plate;

[0035] The substrate is processed according to the CAD model, and at least two functional areas are formed on the substrate surface, including AFL planar area, periodic height modulation pattern area, stepped height area, two-dimensional pattern area, morphological fidelity area and auxiliary visual texture area.

[0036] Laser microtextures with a preset period and shape are processed in the functional area, visual marks are made in the selected area, and corresponding surface treatments are performed to form the desired surface texture in the corresponding functional area.

[0037] The three-dimensional calibration plate is measured using a measuring device to obtain measurement reference data of its surface; the measurement reference data is used to construct a reference measurement model.

[0038] This invention integrates multiple functional regions defined by the same CAD model on a single substrate, enabling simultaneous and efficient evaluation of various performance indicators such as system depth scale, lateral positioning, planar noise, and complex surface reconstruction. This avoids the repetitive clamping and coordinate system transformation errors caused by using multiple independent standard components. The pre-set surface textures in each region effectively alleviate the matching difficulties of highly reflective or smooth surfaces in binocular reconstruction, enhancing feature visibility. The multi-dimensional evaluation results are transformed into clear and comparable physical indicators, supporting quantitative debugging and comparison of imaging algorithms. The 3D calibration board provided by this invention can serve as a standardized foundational tool throughout the entire process of stereoscopic microscopy system development, factory acceptance, and long-term stability monitoring, improving the calibration capabilities of the microscopy system. Attached Figure Description

[0039] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0040] Figure 1 This is an example of a three-dimensional calibration plate of the present invention;

[0041] Figure 2 This is a schematic diagram of an example star-shaped multi-period grating region of the present invention;

[0042] Figure 3 This is a schematic diagram illustrating an example of the design of the step height and reference height according to the present invention;

[0043] Figure 4 This is a schematic diagram of a 6×6 cross-shaped grid area as an example of the present invention;

[0044] Figure 5 This is a schematic diagram illustrating the fidelity evaluation of a sinusoidal surface morphology as an example of the present invention.

[0045] Figure 6 This is a schematic diagram of the structure of a three-dimensional calibration plate containing lattice, stripes and multi-morphological regions, as an example of the present invention. Detailed Implementation

[0046] To make the technical problems solved, the technical solutions, and the beneficial effects of the present invention clearer, the present invention will be further described in detail below with reference to embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present invention and are not intended to limit the present invention.

[0047] This invention provides a three-dimensional calibration plate for a stereoscopic microscopy imaging system, the three-dimensional calibration plate having at least two functional areas;

[0048] Each of the aforementioned functional areas has a specific three-dimensional geometry and surface texture;

[0049] All the aforementioned functional areas are defined using the same CAD model;

[0050] Each of the aforementioned functional areas is used to collect specific imaging evaluation metrics.

[0051] Understandably, this embodiment provides a three-dimensional calibration plate for a stereomicroscopic imaging system. The main body of the three-dimensional calibration plate can be a square substrate with a unified coordinate reference. Preferably, the substrate size is designed to be approximately 50mm × 50mm. The substrate material can be selected according to specific processing technology and imaging band requirements, such as metal, silicon, glass, ceramic, etc. On the substrate surface, at least two functional areas with different specific three-dimensional geometric shapes and surface textures can be precisely divided according to the same pre-designed CAD model. Based on the same CAD model, strict alignment of all functional areas in spatial coordinates is ensured, providing a high-precision geometric reference for subsequent multi-position and multi-angle measurements. It should be noted that, here, the stereomicroscopic imaging system is a non-contact optical measurement system for three-dimensional morphology measurement of microscopic objects. This stereomicroscopic imaging system can simultaneously acquire two-dimensional image information of objects at the microscopic scale (such as at objective lens magnification) and calculate the three-dimensional spatial coordinates of each point on the object surface through a three-dimensional reconstruction algorithm, thereby achieving high-precision quantification and analysis of the geometric features such as morphology, size, and contour of microscopic objects.

[0052] Optionally, the functional area can be any one of the following: an AFL planar area (ideal plane fitting plane area), a periodic height modulation pattern area, a stepped height area, a two-dimensional pattern area, a shape fidelity area, or an auxiliary visual texture area.

[0053] Understandably, functional areas include, but are not limited to, AFL planar areas, periodic height-modulated pattern areas, stepped height areas, two-dimensional pattern areas, morphological fidelity areas, and auxiliary visual texture areas. Different functional areas have specific three-dimensional geometric shapes.

[0054] The periodic height-modulated pattern region is a functional area whose surface morphology is modulated by creating microstructures with periodic height variations on the substrate surface. This region generates quantifiable spatial frequency signals during imaging, which are used to evaluate system performance such as resolution and contrast transfer function. Typical patterns include star-shaped gratings, linear variable-period gratings, concentric ring gratings, checkerboard height-modulated patterns, and multi-directional periodic arrays.

[0055] The stepped height region contains a series of stepped structures with known height differences. The step spacing can be designed with equal, non-equal, or logarithmic intervals to adapt to the measurement needs and error modeling of different depth ranges.

[0056] Two-dimensional patterned regions refer to planar functional areas formed on the surface of a substrate, consisting of specific geometric shapes and arrangement rules. These regions provide reliable corresponding points or feature references for stereo vision systems through high-contrast, precisely positioned two-dimensional patterns (such as cross grids, dot arrays, AprilTag / ArUco type markers, or coded dot matrices). They are primarily used to evaluate performance indicators such as lateral (xy-direction) mapping deviation, positioning error, and feature matching stability.

[0057] The morphology fidelity region contains irregular three-dimensional morphologies used to evaluate the system's ability to reconstruct complex surfaces. These morphologies can include sinusoidal surfaces, cylinders / semicircles, triangular ridges, and can be extended to spheres, saddle surfaces, random rough surfaces, microgroove arrays, or other features similar to the morphology of the actual object being measured.

[0058] The auxiliary visual texture region is processed with random dots, diagonal stripes or other visual markers in the selected area to enhance feature visibility and test the matching robustness of the system under different lighting and viewing angles.

[0059] By flexibly selecting and combining the above functional areas, calibration plates can be customized for specific evaluation tasks, enabling precise quantitative evaluation of the multidimensional performance of stereomicroscopic imaging systems.

[0060] like Figure 1 As shown, Figure 1 As an example, a three-dimensional calibration board includes the following functional areas: a sinusoidal wave region (a specific manifestation of a shape fidelity region) with a periodic wavy surface shape; an AFL plane region with an optically flat reference plane shape; a two-dimensional dot matrix region (a specific manifestation of a two-dimensional pattern region) with a regularly arrayed dot-like recessed shape; a concentric cylinder region (a specific manifestation of a shape fidelity region) with a coaxial rotating body and a circular arc contour shape; a periodic height modulation pattern region with a radially distributed gradient fan-shaped stripe shape; and a stepped height region with a stepped geometry with discrete height differences.

[0061] For specific applications in different regions, laser processing technology is used to create micron-level surface textures on the substrate surface. For example, triangular grooves or pyramid-like structures with a period of approximately 20 μm are processed in designated areas. These customized surface textures not only effectively improve local contrast during microscopic imaging but also significantly reduce the interference of highly reflective surfaces on stereo matching algorithms, thus enabling each geometric region to simultaneously possess the dual attributes of "measurable geometric truth" and "matchable texture." In some examples, to meet diverse evaluation needs, certain functional areas can also be processed with auxiliary visual textures such as random dot matrix or diagonal stripes, or the substrate surface can be treated with specific processes such as hydrophobicity or matte finish.

[0062] Optionally, the AFL planar region is used to acquire imaging evaluation indicators including flatness deviation;

[0063] The periodic highly modulated pattern region is used to acquire imaging evaluation metrics including topographic spatial resolution.

[0064] The imaging evaluation indicators used for the stepped height region include depth magnification factor and linear deviation.

[0065] The two-dimensional pattern region is used to collect imaging evaluation indicators, including xy mapping deviation and lateral positioning error.

[0066] The morphology fidelity region is used to collect imaging evaluation indicators, including morphology fidelity.

[0067] The auxiliary visual texture region is used to collect imaging evaluation metrics including stereo matching stability and local feature visibility.

[0068] Understandably, each functional area is designed for a specific performance dimension of the stereomicroscopy imaging system, and is used to collect corresponding imaging evaluation indicators during system calibration and performance evaluation.

[0069] Specifically, the AFL planar region is mainly used to evaluate the flatness deviation and noise distribution characteristics of the system. The specific method is as follows: first, based on the reconstructed point cloud data within this region, an ideal plane equation is fitted. Then, the height residuals from each sampling point to the fitting plane are calculated: Flatness deviation can be quantified using statistics such as the peak-to-valley value (PV), root mean square value (RMS), standard deviation, and the proportion of points exceeding the threshold. This area can also be used to observe the system's noise characteristics in the depth direction, local fringe artifacts, and error distribution at the edge of the field of view, thereby comprehensively evaluating the overall measurement accuracy and repeatability of the system.

[0070] Periodic highly modulated pattern regions are used to evaluate the system's topographic spatial resolution. The ASP star-shaped grating region is a typical special case of a periodic highly modulated pattern region. This region achieves continuous variation of the grating period along the radial or sector direction through its star-shaped arrangement. During evaluation, the reconstructed trench depth or modulation amplitude is extracted for each period position. and with known reference depth Comparison. When When the resolution drops to 50%, the corresponding spatial period p50 can be defined as the limit of the system's topographic spatial resolution in that direction. This design enables simultaneous evaluation of the system's reconstruction capabilities at different directions and spatial frequencies within a single structure. For example... Figure 2 As shown, Figure 2 This is a schematic diagram of a star-shaped multi-period grating region. The grating region includes wedge-shaped fringes (rays) radiating outward from the center. These fringes are periodically and uniformly arranged around the central point and are surrounded by a circular boundary.

[0071] In the stepped height region, using a stepped structure with known height differences, the system's depth magnification factor and linear deviation are determined. Specifically, in this region, the system measures height. Compared with known reference height A linear response relationship can be established between them: Among them, the regression slope Indicates the depth magnification factor, intercept This represents the zero-point deviation. It can be obtained through linear fitting. and Furthermore, the maximum local deviation of each step measurement point relative to the fitted straight line is calculated: This value is the linearity deviation, used to assess the accuracy of the system's depth scale and the presence of nonlinear response, thereby verifying the accuracy of the system's height measurement at different depths of field. For example... Figure 3 As shown, Figure 3 This diagram illustrates the design of the step height and reference height. The upper and lower parts of the diagram represent two different three-dimensional perspectives (convex and concave), showcasing a multi-level grayscale step distribution from pure black to pure white. The height change of each step corresponds to a specific grayscale reference value, clearly expressing the hierarchical structure of height quantification.

[0072] A two-dimensional patterned region is used to evaluate the system's xy-mapping deviation and lateral positioning error. Within this region (e.g., a two-dimensional cross-shaped grid region), the measured coordinates of each grid intersection or cross center are obtained through sub-pixel precision feature extraction. and with known reference coordinates By comparison, the two-dimensional error components at each point are obtained: , Based on all error data, the global RMS error, maximum error, and systematic distortion trends along the x and y directions can be calculated. The error vector, statistical error, and spatial distribution patterns output in this region can quantitatively characterize the system's field-of-view distortion and lateral positioning accuracy. Figure 4 As shown, Figure 4 This is a schematic diagram of a 6×6 cross-shaped grid area. This grid area is composed of multiple square grooved units with chamfered features arranged periodically.

[0073] The morphology fidelity region is used to comprehensively evaluate the system's ability to reconstruct complex micro-morphologies. This region includes typical geometric features such as sinusoidal surfaces, cylindrical surfaces, semicircles, and triangular ridges, allowing for systematic testing of the algorithm's detail preservation and reconstruction fidelity for continuous surfaces, sharp edges, local peaks, and high curvature regions. During evaluation, the reconstructed micro-morphology is compared with the ground truth CAD model, extracting indicators including root mean square (RMS) morphology error, maximum contour error, amplitude error, phase error, edge transition width, and curvature error, thereby quantifying the system's overall performance under complex morphologies. This region overcomes the limitations of evaluating simple structures such as single planes, steps, or lattices, providing crucial evaluation criteria for the system's morphology reconstruction capabilities in real-world complex scenes. Figure 5 As shown, Figure 5 A schematic diagram illustrating the fidelity evaluation of sinusoidal surface topography is presented, which is achieved by superimposing a sinusoidal curve onto a three-dimensional undulating gray base surface.

[0074] The auxiliary visual texture area is mainly used to monitor the stability of stereo matching and the visibility of local features. By introducing random textures with specific reflective properties or low contrast, it collects indicators such as matching success rate, disparity stability, and sensitivity to occluded or highly reflective areas, ensuring that the system can still obtain robust and continuous disparity maps under complex working conditions.

[0075] This embodiment integrates multiple functional areas defined by the same CAD model on a single substrate, enabling simultaneous and efficient evaluation of various performance indicators such as system depth scale, lateral positioning, planar noise, and complex surface reconstruction. This avoids the repetitive clamping and coordinate system transformation errors caused by using multiple independent standard parts. The pre-set surface textures in each area effectively alleviate the matching difficulties of highly reflective or smooth surfaces in binocular reconstruction, enhancing feature visibility. The multi-dimensional evaluation results are transformed into clear and comparable physical indicators, supporting quantitative debugging and comparison of imaging algorithms. Therefore, the 3D calibration board provided in this embodiment can serve as a standardized basic tool, applicable throughout the entire process of research and development, factory acceptance, and long-term stability monitoring of stereoscopic microscopy imaging systems.

[0076] Optionally, the three-dimensional geometry of the AFL planar region includes a plane;

[0077] The three-dimensional geometry of the periodic highly modulated pattern region includes planar, radial star, multi-sector, multi-period groove, or stepped grating.

[0078] The three-dimensional geometry of the stepped height region includes a step structure that varies along the height direction;

[0079] The three-dimensional geometry of the two-dimensional pattern area includes 6×6 cross or grid markings;

[0080] The three-dimensional geometry of the morphological fidelity region includes sinusoidal surfaces, cylinders / semicircles, and / or triangular ridges;

[0081] The three-dimensional geometry of the auxiliary visual texture region includes random dots and diagonal stripes.

[0082] Understandably, the multiple functional areas of the 3D calibration plate are defined and manufactured in an integrated manner through the same CAD model, ensuring that each area has a unified and accurate spatial coordinate reference.

[0083] Specifically, the AFL planar region is designed as a large-area optically flat reference plane, the surface of which can be selectively covered with micro-textures or matte finishes to suppress specular reflections and accurately reflect the system's flatness deviations and noise distribution. The periodic highly modulated pattern region combines a planar substrate with a radial star structure, internally integrating multi-sector and multi-period groove gratings. Its periodicity changes regularly from the center outwards, and its depth is a known fixed value, specifically used to determine the system's topographic spatial resolution.

[0084] In addition, to comprehensively examine the system's depth perception and spatial positioning capabilities, the calibration board also integrates a stepped height region and a two-dimensional pattern region. The stepped height region features stepped structures with known discrete differences along the height direction; the example parameter range can be designed to be -1.00mm to +1.00mm, with a grading accuracy of 0.10mm / level, used to calibrate the depth magnification factor. For systems with a small field of view or shallow depth range, the height range can be reduced or the number of steps decreased; for systems with a large depth of field, the height range can be expanded. The two-dimensional pattern region uses a regular array of cross or grid markers, with a recommended marker width of approximately 1.5mm and a feature spacing of approximately 3.5mm, facilitating high-precision xy mapping deviation analysis. The morphology fidelity region constructs complex micro-geometry such as sinusoidal surfaces, cylinders / semicircles, or triangular roof ridges; its period and amplitude can be flexibly designed according to the actual field of view and resolution, quantitatively detecting the fidelity of the reconstructed complex morphology using algorithms.

[0085] Finally, the auxiliary visual texture area features a random dot matrix and diagonal stripes. In some examples, the random dot matrix is ​​designed to be approximately 0.5 mm in diameter and randomly distributed. The diagonal stripes are approximately 0.5 mm wide and are positioned at +45° and -45° angles on the upper surface or slope to distinguish different facets and enhance matching robustness. This area effectively simulates low-texture or high-reflectivity interference in industrial environments, and is used to comprehensively analyze the impact of factors such as occlusion and reflection on stereo matching stability. Figure 6 As shown, Figure 6 This is a schematic diagram of the structure of a three-dimensional calibration plate containing lattice, stripe, and multi-morphological regions.

[0086] The three-dimensional calibration plate provided in this embodiment achieves accurate quantitative evaluation of the multi-dimensional performance of the microscopic vision system through the synergistic effect of specific geometric shapes and recommended parameters in each region.

[0087] Optionally, the surface texture includes micron-scale surface microstructures; the surface microstructures are one or more of the following: triangular grooves, pyramidal structures, cones, pyramids, frustums, hexagonal pits, parallel grooves, and vertical grooves; and / or,

[0088] The surface texture is formed by laser processing.

[0089] Understandably, the micron-scale surface microstructures of the functional area can employ triangular trenches or pyramid-like structures with a period of approximately 20 μm and a feature height or trench depth of approximately 20 μm. This microtexture can effectively enhance the local feature contrast of the imaging area and reduce the interference of highly reflective surfaces on stereo matching, thereby enabling the same geometric area to simultaneously possess the dual attributes of "measurable geometric truth" and "matchable texture." As an alternative, periodic microstructures such as cones, pyramids, frustums, hexagonal pits, parallel trenches, and vertical trenches can also be used, for example, an array with a structural size of approximately 30 μm × 40 μm and a period of approximately 30 μm. The specific orientation, period, height, and duty cycle of the microtexture can be adaptively adjusted according to the reflectivity of the substrate material, the lens magnification of the imaging system, and the camera pixel size.

[0090] Preferably, the surface microstructure is prepared by laser processing. Laser processing has high precision and good controllability, and can accurately and consistently form the above-mentioned micron-level surface textures, and facilitates the flexible design and preparation of complex structural patterns.

[0091] If the calibration plate is subjected to surface treatments such as hydrophobic, anti-fouling, or anti-reflective treatments after laser processing, it is necessary to ensure that the coating thickness and uniformity do not compromise its reference geometric accuracy, or to re-perform high-precision measurements after the treatment to update the reference data.

[0092] This invention also provides a performance evaluation method for a stereomicroscopic imaging system, comprising:

[0093] Image data was acquired using the stereomicroscopic imaging system to be evaluated for any of the three-dimensional calibration plates mentioned above, and image measurement data was obtained.

[0094] The image measurement data is registered with the reference data to obtain registration measurement data;

[0095] The topographic data of each functional region is extracted from the registration measurement data, and the imaging evaluation index corresponding to the functional region is calculated based on the topographic data.

[0096] Understandably, the 3D calibration plate obtained in the previous embodiment is placed within the effective field of view and depth of field of the stereo microscopy imaging system to be evaluated. At different depths, multiple lateral positions, and various tilt angles on the calibration plate, a series of high-quality image data are acquired through a stereo vision system (such as dual cameras). Using the calibration and 3D reconstruction algorithms built into the system to be evaluated, stereo matching and 3D reconstruction are performed on the acquired images to obtain image measurement data such as 3D point clouds, depth maps, or mesh models of the calibration plate surface. If a compensation model needs to be established, the calibration plate can be moved within the effective depth of field, and the acquisition and calculation can be repeated to obtain the error distribution that varies with depth or field of view position.

[0097] Subsequently, the reconstructed 3D measurement data is rigidly or similarly registered with the nominal CAD model of the calibration plate or the reference measurement model (i.e., baseline data) obtained by high-precision metrology equipment to eliminate pose differences in the global coordinate system and obtain high-precision registered measurement data. After registration, local topographic data corresponding to planar regions, stepped regions, 2D meshes, star-shaped gratings, and complex topographic regions are extracted from the overall data according to the preset geometric boundaries of each functional area.

[0098] For the topographic data extracted from each functional region, a specially designed imaging evaluation index is calculated. For example, flatness deviation and noise distribution are calculated based on AFL planar region data, depth magnification factor and linearity deviation are calculated based on stepped height region data, spatial resolution (p50) is calculated based on periodic height modulation pattern region data, and xy mapping deviation is calculated based on two-dimensional pattern region data.

[0099] This embodiment achieves integrated, efficient evaluation of the multidimensional performance of a stereomicroscopic imaging system by integrating a calibration plate with multiple feature functional areas and a standardized quantitative evaluation process; it transforms the system's core indicators, such as flatness, depth accuracy, spatial resolution, and ability to reconstruct complex morphologies, into a series of objective and comparable specific values.

[0100] Optionally, the reference data includes a CAD model or a reference measurement model.

[0101] Understandably, the reference data can be the nominal CAD model of the calibration board or a reference measurement model that has undergone high-precision metrology. When using the nominal CAD model, it serves as an ideal design benchmark, suitable for quickly assessing the system's reconstruction capability and deviation from the theoretical design in scenarios where the requirements for absolute accuracy are relatively broad, such as initial system debugging or when the requirements for absolute accuracy are relatively broad. When using the reference measurement model, it essentially consists of high-precision data such as point clouds and meshes obtained by measuring the physical calibration board using equipment such as coordinate measuring machines and white-light interferometers. It contains the actual geometric details and microstructures generated during the manufacturing process, and therefore can serve as a more accurate "physical truth," suitable for scenarios such as final system acceptance, high-precision error calibration, or rigorous comparison between different systems. The choice of reference data can be flexibly determined based on the evaluation stage and accuracy requirements.

[0102] Optionally, the imaging evaluation metrics include topographic spatial resolution, depth magnification factor, linearity deviation, xy mapping deviation, lateral positioning error, topographic fidelity, stereo matching stability, and local feature visibility.

[0103] Understandably, imaging evaluation metrics are a set of key performance dimensions of a system, such as topographic spatial resolution, depth magnification factor, linearity deviation, xy mapping deviation, lateral positioning error, topographic fidelity, stereo matching stability, and local feature visibility.

[0104] Specifically, the spatial resolution of the morphology and the depth magnification factor can be calculated using periodic height-modulated pattern regions and stepped height regions, respectively. The former quantifies the system's ability to resolve fine structures using the limiting spatial period (e.g., p50), while the latter characterizes the depth measurement scale using the linear regression slope. The xy mapping deviation and lateral positioning error are obtained by analyzing the reprojection error of feature points in the two-dimensional pattern region, used to quantify the system's lateral distortion and positioning accuracy. Morphology fidelity is calculated by comparing the reconstructed data of complex morphological regions (e.g., sine waves, cylinders) with baseline data to evaluate the system's ability to reconstruct complex surfaces. Stereo matching stability and local feature visibility are evaluated by analyzing the matching success rate and feature extraction consistency of the auxiliary visual texture region under different illuminations and viewing angles, reflecting the system's robustness under actual working conditions. These indicators together constitute a complete and quantitative evaluation system for the performance of the stereo microscopy imaging system.

[0105] Optionally, after extracting the topographic data of each functional region from the registration measurement data and calculating the imaging evaluation index corresponding to the functional region based on the topographic data, the method further includes:

[0106] A comprehensive performance report is generated by statistically analyzing imaging evaluation indicators from multiple dimensions and under multiple conditions.

[0107] Understandably, various evaluation indicators calculated from multiple locations, perspectives, and even algorithm versions can be summarized, statistically analyzed, and compiled into a comprehensive performance report. This report can be directly used for system acceptance testing, objective comparison of different algorithms or systems, and provides a precise data foundation for subsequent error compensation modeling.

[0108] This invention also provides a method for manufacturing any of the three-dimensional calibration plates described above, comprising:

[0109] Based on the parameters and requirements of the stereomicroscopic imaging system to be evaluated, design the CAD model of the three-dimensional calibration plate;

[0110] The substrate is processed according to the CAD model, and at least two functional areas are formed on the substrate surface, including AFL planar area, periodic height modulation pattern area, stepped height area, two-dimensional pattern area, morphological fidelity area and auxiliary visual texture area.

[0111] Laser microtextures with a preset period and shape are processed in the functional area, visual marks are made in the selected area, and corresponding surface treatments are performed to form the desired surface texture in the corresponding functional area.

[0112] The three-dimensional calibration plate is measured using a measuring device to obtain measurement reference data of its surface; the measurement reference data is used to construct a reference measurement model.

[0113] Understandably, the first step is to design a CAD model of the 3D calibration board based on the field of view, depth of field, pixel size, and algorithm requirements of the stereomicroscopic imaging system to be evaluated. The model is typically designed to be approximately 50mm × 50mm, and it must define the precise 3D geometry and spatial layout of at least two functional areas: the AFL planar region, the periodic height modulation pattern region, the stepped height region, the 2D pattern region, the morphology fidelity region, and the auxiliary visual texture region. All regions originate from the same CAD model to ensure a unified design benchmark.

[0114] Next, based on the imaging wavelength and processing method, suitable materials such as metal, silicon, or glass are selected as the substrate. According to the CAD model, the three-dimensional geometric structures of each functional area are sequentially formed on the substrate surface using precision machining techniques, such as planes, steps, grids, gratings, and complex curved surfaces. Based on this, laser micro-texturing is applied to designated areas, for example, forming triangular grooves or pyramid-like structures with a period of approximately 20 μm, or quadrilateral microstructures with a period of approximately 30 μm, thereby imparting fine textures to the geometric surface suitable for stereo matching.

[0115] Subsequently, random dots, diagonal stripes, and other visual markings are processed on selected areas (such as areas for auxiliary visual texture). Stripes with opposite directions can be used on different bevels to enhance identification. Depending on the specific application environment, the entire calibration plate can undergo hydrophobic, matte, or low-reflection surface treatments to optimize its optical properties. If the surface treatment may affect the geometric dimensions, the final measurement can only be performed after this step.

[0116] Finally, high-precision measuring equipment such as coordinate measuring machines, laser scanners, or white-light interferometers are used to perform full-size precision measurements on the manufactured calibration plate and the surface treated (if applicable). The acquired high-density point cloud or mesh data is used to construct a unique reference measurement model for the calibration plate, accurately representing its actual geometric state. This reference measurement model, together with the CAD model, serves as a deliverable, providing reliable benchmark data for subsequent system performance evaluation.

[0117] This embodiment integrates multiple functional areas with specific geometric shapes and surface textures on a single substrate through integrated CAD design and manufacturing processes. This achieves the goal of simultaneously evaluating multiple performance indicators of the system using the same calibration component, avoiding clamping errors introduced by multiple standard components. Laser micro-textures and visual markings are processed on the functional areas, enabling the high-precision geometric surfaces to possess the dual attributes of "measurable geometric truth value" and "matchable texture," effectively solving the problem of stereo matching of highly reflective or weakly textured surfaces in microscopic vision. By establishing a unique reference measurement model based on the physical object through high-precision measurement, an authoritative benchmark based on the manufactured object is provided for system evaluation, ensuring the accuracy and reliability of the evaluation results.

[0118] 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 three-dimensional calibration plate for a stereomicroscopic imaging system, characterized in that, The three-dimensional calibration board is provided with at least two functional areas; Each of the aforementioned functional areas has a specific three-dimensional geometry and surface texture; All the aforementioned functional areas are defined using the same CAD model; Each of the aforementioned functional areas is used to collect specific imaging evaluation metrics.

2. The three-dimensional calibration plate for a stereoscopic microscopic imaging system as described in claim 1, characterized in that, The functional area can be any one of the following: AFL planar area, periodic height modulation pattern area, stepped height area, two-dimensional pattern area, shape fidelity area, and auxiliary visual texture area.

3. The three-dimensional calibration plate for a stereoscopic microscopic imaging system as described in claim 2, characterized in that, The flatness deviation is one of the imaging evaluation metrics used to acquire images in the AFL planar region. The periodic highly modulated pattern region is used to acquire imaging evaluation metrics including topographic spatial resolution. The imaging evaluation indicators used for the stepped height region include depth magnification factor and linear deviation. The two-dimensional pattern region is used to collect imaging evaluation indicators, including xy mapping deviation and lateral positioning error. The morphology fidelity region is used to collect imaging evaluation indicators, including morphology fidelity. The auxiliary visual texture region is used to collect imaging evaluation metrics including stereo matching stability and local feature visibility.

4. The three-dimensional calibration plate for a stereoscopic microscopic imaging system as described in claim 2, characterized in that, The three-dimensional geometry of the AFL planar region includes a plane; The three-dimensional geometry of the periodic highly modulated pattern region includes planar, radial star, multi-sector, multi-period groove, or stepped grating. The three-dimensional geometry of the stepped height region includes a step structure that varies along the height direction; The three-dimensional geometry of the two-dimensional pattern area includes 6×6 cross or grid markings; The three-dimensional geometry of the morphological fidelity region includes sinusoidal surfaces, cylinders / semicircles, and / or triangular ridges; The three-dimensional geometry of the auxiliary visual texture region includes random dots and diagonal stripes.

5. The three-dimensional calibration plate for a stereoscopic microscopic imaging system as described in claim 1, characterized in that, The surface texture includes micron-scale surface microstructures; the surface microstructures are one or more of the following: triangular grooves, pyramid-like structures, cones, pyramids, frustums, hexagonal pits, parallel grooves, and vertical grooves; and / or, The surface texture is formed by laser processing.

6. A performance evaluation method for a stereomicroscopic imaging system, characterized in that, include: Image data was acquired using the stereomicroscopic imaging system to be evaluated on the three-dimensional calibration plate described in any one of claims 1 to 5 to obtain image measurement data. The image measurement data is registered with the reference data to obtain registration measurement data; The topographic data of each functional region is extracted from the registration measurement data, and the imaging evaluation index corresponding to the functional region is calculated based on the topographic data.

7. The performance evaluation method for the stereomicroscopic imaging system as described in claim 6, characterized in that, The reference data includes CAD models or reference measurement models.

8. The performance evaluation method for the stereomicroscopic imaging system as described in claim 6, characterized in that, The imaging evaluation metrics include topographic spatial resolution, depth magnification factor, linearity deviation, xy mapping deviation, lateral positioning error, topographic fidelity, stereo matching stability, and local feature visibility.

9. The performance evaluation method for the stereomicroscopic imaging system as described in claim 6, characterized in that, After extracting the topographic data of each functional region from the registration measurement data and calculating the imaging evaluation index corresponding to the functional region based on the topographic data, the method further includes: A comprehensive performance report is generated by statistically analyzing imaging evaluation indicators from multiple dimensions and under multiple conditions.

10. A method for manufacturing a three-dimensional calibration plate as described in any one of claims 1 to 5, characterized in that, include: Based on the parameters and requirements of the stereomicroscopic imaging system to be evaluated, design the CAD model of the three-dimensional calibration plate; The substrate is processed according to the CAD model, and at least two functional areas are formed on the substrate surface, including AFL planar area, periodic height modulation pattern area, stepped height area, two-dimensional pattern area, morphological fidelity area and auxiliary visual texture area. Laser microtextures with a preset period and shape are processed in the functional area, visual marks are made in the selected area, and corresponding surface treatments are performed to form the desired surface texture in the corresponding functional area. The three-dimensional calibration plate is measured using a measuring device to obtain measurement reference data of its surface; The measurement reference data is used to construct a reference measurement model.

Images