Calibration method for absolute accuracy of hidden point centroid based on three-coordinate instrument

Abstract

The application claims a kind of hidden point centroid absolute accuracy calibration method based on three coordinate instrument, belongs to the technical field of terahertz reflection imaging centroid positioning and accuracy calibration, it includes the following steps: preparation of calibration carrier with high-low step structure and layout terahertz reflection label;All labels are measured by non-contact three coordinate measuring instrument, and the absolute true value coordinates of the three-dimensional centroid of each label are obtained;The calibration carrier is scanned and imaged by using the terahertz reflection imaging system, and the three-dimensional centroid coordinates of each label are solved by image processing;With the true value of centroid output by three coordinates as the reference, the interplane distance and height difference are compared respectively, and the absolute accuracy calibration of three-dimensional centroid of terahertz imaging system hidden point is completed.

Description

Technical Field

[0001] This invention belongs to the field of terahertz reflective imaging centroid positioning and accuracy calibration technology, specifically a method for absolute accuracy calibration of the centroid of hidden points based on a coordinate measuring machine. Background Technology

[0002] Terahertz imaging technology, with its advantages of good penetration, high resolution, non-contact operation, and no ionizing radiation, is widely used in non-destructive testing, structural identification, and target localization. Reflective terahertz imaging is the mainstream technology for detecting surface features of complex components and identifying concealed targets. In practical engineering applications, the absolute accuracy of the centroid localization of target points and concealed marker points by the terahertz imaging system is a core indicator for evaluating the system's detection capabilities and ensuring the reliability of measurement results. Therefore, calibrating the centroid localization accuracy of the terahertz imaging system has become an indispensable step in this field.

[0003] Currently, the industry mainly uses two types of calibration methods for terahertz imaging systems: The first is to use planar targets such as standard checkerboards and standard scale plates for calibration. This method can only complete pixel correction and relative positioning error analysis in a two-dimensional plane, and cannot simulate three-dimensional spatial positioning scenarios in actual working conditions with multi-layered components and staggered structures. It is also difficult to effectively calibrate hidden points and feature points with varying heights and levels, resulting in a significant deviation between the calibration scenario and actual application. The second method uses ordinary metal blocks in conjunction with traditional optical imaging for comparison and calibration. However, traditional optical imaging is easily affected by illumination, surface reflection, and obstructions. For small-sized reflective tags and long-focal-length terahertz light spots, there are problems such as difficulty in edge extraction, contour breakage, and large positioning deviations. Moreover, most conventional calibration methods can only obtain relative accuracy, making it difficult to obtain absolute centroid coordinates that can be used as a reference. Ultimately, this results in the inability to accurately quantify the absolute accuracy of the three-dimensional centroid of the terahertz imaging system.

[0004] Meanwhile, existing calibration schemes have significant shortcomings in the measurement of tiny marker points: the reflective tags used in terahertz imaging are typically only millimeters in size, and traditional contact measurement tools are prone to causing tag deformation and displacement, further introducing measurement errors; while traditional image processing often relies on gradient-based edge detection algorithms such as Canny, Sobel, and Prewitt, which are prone to edge loss and region segmentation failure when faced with long focal diffused light spots and signal noise interference in terahertz imaging, and cannot achieve sub-pixel-level centroid solving.

[0005] This invention proposes a method based on a coordinate measuring machine to obtain the three-dimensional absolute centroid coordinates of a tiny reflective tag using a high-precision metrology device as the true value benchmark. Combined with terahertz reflection imaging, dedicated image segmentation, and centroid solving algorithms, the absolute accuracy calibration of the three-dimensional centroids of hidden points and marker points under high and low stepped structures is completed.

[0006] Existing calibration schemes mostly only analyze errors at points within a single plane, rarely considering testing in multi-level staircases or other complex three-dimensional structures, resulting in low compatibility with real-world, complex working conditions. There is currently no mature solution that utilizes a coordinate measuring machine to obtain the three-dimensional absolute centroid coordinates of the miniature reflective tag as a benchmark, and then combines this with a terahertz imaging system to achieve accurate calibration of tag points in multi-level staircase structures. Therefore, it is difficult to comprehensively and accurately calibrate the absolute accuracy of the centroid of hidden points in a terahertz imaging system. Summary of the Invention

[0007] This invention aims to solve the problems of the prior art mentioned above. It proposes a method for calibrating the absolute accuracy of the centroid of hidden points based on a coordinate measuring machine. The technical solution of this invention is as follows:

[0008] A method for calibrating the absolute accuracy of the centroid of a hidden point based on a coordinate measuring machine includes the following steps:

[0009] A calibration carrier with a stepped structure was prepared and terahertz reflective tags were deployed.

[0010] All labels were measured non-contactly using a coordinate measuring machine to obtain the absolute true coordinates of the three-dimensional centroid of each label.

[0011] A terahertz reflection imaging system was used to scan and image the calibration carrier, and the three-dimensional centroid coordinates of each label were solved by image processing.

[0012] Using the true centroid value output by the three coordinate system as a reference, the absolute accuracy calibration of the three-dimensional centroid of the hidden point of the terahertz imaging system is completed by comparing the plane spacing and height difference.

[0013] Furthermore, the preparation of the calibration carrier with a stepped structure and the deployment of terahertz reflective tags specifically includes:

[0014] Preparation of calibration carrier and micro terahertz reflective tags: Two-stage aluminum metal stepped blocks are fabricated as calibration carriers. Each stepped block comprises two steps with heights of 10mm and 20mm, respectively, and each step is 50mm long and 100mm wide. A three-layer terahertz reflective tag with a size of 2mm × 2mm is fabricated. From top to bottom, the tag consists of a gold reflective layer, a chromium adhesive layer, and a silicon dioxide substrate layer. Three tags are evenly adhered to the surface of the first step and three tags are evenly adhered to the surface of the second step using aviation adhesive.

[0015] Furthermore, S2, obtaining the true value of the label's three-dimensional absolute centroid based on a coordinate measuring machine, specifically includes:

[0016] S201. Fix the metal stepped block with the label to the worktable of the coordinate measuring machine. Establish the workpiece coordinate system O-XYZ with the bottom surface of the stepped block as the reference surface and the two mutually perpendicular side surfaces as auxiliary references. Select a non-contact optical probe with a spot diameter of no more than 0.4 mm and a lateral resolution of ≥5 μm.

[0017] S202. Perform measurements on 6 tags sequentially: Identify the complete outline of the tag by image binarization and laser reflection intensity threshold segmentation, and collect no less than 100 outline / surface point clouds by scanning with a step spacing of ≤0.05mm; project the point cloud onto the step fitting plane, fit the circumscribed rectangle of the tag using the least squares method, and calculate the geometric center to obtain the single centroid coordinates; repeat the measurement 3 times for each tag, and take the average of the three coordinates as the true value Pi of the three-dimensional absolute centroid of the tag, i=1~6;

[0018] S203. Based on the spatial two-point distance formula, calculate the standard distance D between the centroids of each label within the same step; and calculate the Z-axis coordinate difference N between the corresponding labels of the two steps as the three-dimensional accuracy calibration benchmark data.

[0019] Furthermore, S3, setting up a terahertz reflective imaging system and acquiring tag imaging data, specifically includes:

[0020] S301. Construct a femtosecond laser-driven reflective terahertz imaging system. The system includes a femtosecond laser, a beam splitter, an optical delay unit, an off-axis parabolic mirror, a photoconductive antenna probe, a stepper motor, and a host computer. A 2.5cm thick heat-insulating tile is laid on the outside of the metal stepped block to simulate the concealed and obstructed working condition. The distance between the probe and the surface of the stepped block is set to 31.6cm.

[0021] S302: Control the stepper motor to drive the terahertz probe to perform XY plane step scanning on the metal stepped block in 1mm steps, collect the terahertz time-domain reflection signal at each position, extract the reflection amplitude to generate the original terahertz grayscale image; calculate the Z-axis imaging coordinate of each tag based on the peak time delay of the time-domain waveform.

[0022] Furthermore, the S4, subpixel-level centroid solution of the terahertz image, specifically includes:

[0023] S401. Perform filtering and interpolation preprocessing on the original terahertz image to suppress noise;

[0024] S402. Use a binarization + distance transform joint region segmentation algorithm to extract the complete closed region of the label;

[0025] S403. The gray-scale weighted average algorithm is used to calculate the sub-pixel level two-dimensional centroid of the segmented closed region. Combined with the time-domain delay Z-axis coordinate, the three-dimensional centroid coordinates Qi of each tag measured by the terahertz system are obtained, i=1~6.

[0026] S404. Calculate the label imaging distance d within the same step and the Z-axis coordinate difference M between the labels corresponding to the two steps.

[0027] Furthermore, the S5, the three-dimensional centroid absolute accuracy calibration and evaluation, specifically includes:

[0028] S501, XY plane accuracy: Compare the standard distance D of the three coordinate system within the same step with the terahertz imaging distance d, the difference is the two-dimensional plane positioning error;

[0029] S502, Z-axis height accuracy: Compare the Z-axis difference N of the three-coordinate reference with the Z-axis difference M of the terahertz imaging. The difference is the positioning error in the depth direction.

[0030] S503, combining two-dimensional and depth errors, completes the absolute accuracy calibration of the three-dimensional centroid of the hidden marker points of the high and low steps in the terahertz imaging system.

[0031] The advantages and beneficial effects of this invention are as follows:

[0032] 1. Construct a high-precision absolute coordinate benchmark, ensuring accurate calibration results.

[0033] This solution uses a coordinate measuring machine with a non-contact optical probe to measure miniature reflective tags and obtain the three-dimensional centroid coordinates of the tags as the absolute true value. This avoids the problem of traditional calibration having no standard reference and only being able to estimate relative errors. Using this as a reference, accuracy comparison can be carried out, which can accurately quantify the positioning error of the terahertz imaging system. The authority and accuracy of the calibration results are greatly improved.

[0034] 2. Adaptable to complex 3D working conditions, with wider coverage of calibration scenarios.

[0035] A three-dimensional test structure with varying heights is constructed using two levels of metal stepped blocks. This structure can simultaneously analyze the distance and coordinate errors of label points on the same plane and at different heights. It breaks through the limitation of traditional calibration, which is only applicable to a single plane. The calibration scenario is more in line with the actual engineering conditions of detecting hidden points with steps and multi-layered components, and the test conclusions are more reliable.

[0036] The innovation of this invention is mainly reflected in the overall method summarized in claim 1, and the specific steps and their synergistic effects corresponding to claims 2, 3, and 5.

[0037] These innovations are not conventional technical means because they solve the technical problem of terahertz imaging systems lacking high-precision, traceable absolute accuracy calibration benchmarks in complex three-dimensional scenes through a series of carefully designed combinations.

[0038] 1. Construct a three-dimensional calibration scene (corresponding to claim 2)

[0039] Innovation: Prepare a calibration carrier (metal step block) with a stepped structure of varying heights, and place terahertz reflection tags on the steps at different heights.

[0040] Overcoming planar limitations: Traditional calibration methods often use two-dimensional planar targets such as checkerboards and scale plates. These planar targets can only calibrate the accuracy in the XY plane, ignoring the accuracy along the longitudinal Z-axis and failing to simulate and evaluate the three-dimensional spatial positioning problems faced by terahertz systems in real-world operating conditions.

[0041] Closely aligned with real-world working conditions: Through its staggered three-dimensional structure, this invention can simultaneously evaluate the planar accuracy (XY axis) and depth accuracy (Z axis) of the calibration system, making the calibration scenario closer to actual applications and the test results more valuable for reference.

[0042] 2. Establish a high-precision absolute truth benchmark (corresponding to claim 3)

[0043] Innovation: A non-contact optical probe of a high-precision coordinate measuring machine is used to scan tiny tags, and the true coordinates of the three-dimensional absolute centroid of each tag are calculated by point cloud fitting.

[0044] Providing an absolute benchmark: Conventional calibration methods mostly only obtain relative accuracy, that is, the relative error between measured values, lacking an authoritative and traceable "true value" as a comparison benchmark. This invention uses the coordinates obtained by a coordinate measuring machine (CMM) of metrology grade as the "absolute true value," providing an indisputable reference standard for subsequent accuracy evaluation.

[0045] Solving the challenge of measuring tiny targets: Traditional contact-based measurements are prone to deformation when measuring millimeter-sized tags. This invention employs a non-contact optical probe, combined with algorithms such as point cloud acquisition and least squares fitting, to achieve high-precision, non-destructive measurement of the centroid of tiny targets—a feat difficult to achieve using conventional methods.

[0046] 3. A dedicated image processing and centroid solving algorithm is employed (corresponding to claim 5).

[0047] Innovative content: Targeting the characteristics of terahertz images, a joint region segmentation algorithm of "binarization + distance transformation" is adopted, and combined with the gray-level weighted averaging method for sub-pixel centroid localization.

[0048] Overcoming the limitations of traditional algorithms: Terahertz imaging suffers from problems such as light spot diffusion and signal noise interference. Traditional gradient-based edge detection algorithms such as Canny and Sobel are prone to edge breakage and region segmentation failure in such images, leading to inaccurate centroid calculation.

[0049] Improving positioning accuracy: This invention abandons the approach of directly performing edge detection and instead adopts a region segmentation-based method. It first extracts the closed region containing the tag completely, and then calculates the centroid. This method is insensitive to noise and edge discontinuities, enabling more stable and accurate sub-pixel-level positioning, thus ensuring the reliability of the terahertz system's measurement results.

[0050] The true inventiveness of this invention lies in combining the above three innovative points to form a complete closed-loop calibration process:

[0051] Three-dimensional carrier (claim 2): proposes the testing requirements for the three-dimensional accuracy of the calibration system.

[0052] Three-coordinate true value (claim 3): provides a high-precision "standard answer" to meet this requirement.

[0053] Dedicated algorithm (claim 5): ensures the accuracy of the measurement results of the terahertz system itself, making the comparison with the "standard answer" meaningful.

[0054] Finally, by comparing the coordinates measured by the terahertz system with the absolute true values ​​provided by the coordinate measuring machine (as described in claims 1 and 6), a precise and comprehensive calibration of the absolute accuracy of the centroid of hidden points in complex 3D scenes using the terahertz imaging system was achieved. This systematic combination, from calibration carrier design and the establishment of a true reference to the application of specialized algorithms, is unprecedented in existing technologies, and therefore possesses outstanding substantive features and significant progress. Attached Figure Description

[0055] Figure 1 This is a schematic diagram of a metal block being processed according to a preferred embodiment of the present invention;

[0056] Figure 2 This is a schematic diagram of a terahertz reflective tag structure;

[0057] Figure 3 The image shows a diagram of a label being affixed to a metal step block.

[0058] Figure 4 A schematic diagram of coordinate measuring machine measurement;

[0059] Figure 5 This is a diagram showing the distribution of labels.

[0060] Figure 6 This is a schematic diagram of a terahertz imaging system.

[0061] Figure 7 This is a schematic diagram of a terahertz probe scanning a metal step.

[0062] Figure 8 Flowchart of image processing and centroid localization algorithm;

[0063] Figure 9 This indicates the images after scanning the centroids of the labels on the two stepped surfaces. Detailed Implementation

[0064] The technical solutions of the embodiments of the present invention will be clearly and thoroughly described below with reference to the accompanying drawings. The described embodiments are merely some embodiments of the present invention.

[0065] The technical solution of the present invention to solve the above-mentioned technical problems is:

[0066] like Figure 1 The diagram shows a metal block being processed. The metal block consists of two steps (hereinafter referred to as a metal step block). The height of the steps is h1 = 20 mm and h2 = 10 mm. The lengths of the steps are l1 = 50 mm and l2 = 50 mm. The width of the steps is w = 100 mm. The material for processing the metal step block is aluminum.

[0067] like Figure 2 The diagram shows the structure of a terahertz reflective tag. The tag has a diameter of 2mm × 2mm and consists of three layers from top to bottom. The top layer is a reflective layer with high reflectivity to terahertz waves and is made of metallic gold. The middle layer is made of metallic chromium and serves as an adhesive. The substrate is made of silicon dioxide, which supports the tag and improves its lifespan.

[0068] like Figure 3 The diagram shows the labels being affixed to the metal step block. Six labels are affixed to the two steps of the metal step block using aviation adhesive, with three metal labels affixed to each step. These labels are used for subsequent centroid positioning of the coordinate measuring machine and the terahertz probe.

[0069] like Figure 4 This is a schematic diagram of coordinate measuring machine measurement. Figure 5 This is a diagram showing the distribution of labels.

[0070] Place the metal stepped block with the terahertz reflection label on the worktable of the coordinate measuring machine (CMM) and fix it in place using a flexible clamp or a special pressure plate to ensure that the stepped block does not shift during the entire measurement process. Use the bottom surface of the metal stepped block as the primary reference plane and its two mutually perpendicular side surfaces as auxiliary reference planes. Use the CMM software to establish a workpiece coordinate system O-XYZ: where the X and Y axes are located in the horizontal plane, and the Z axis is perpendicular to the worktable surface and upwards.

[0071] Since the label size is only 2mm × 2mm, a non-contact optical probe is used to avoid label deformation or displacement caused by contact force. The spot diameter of the selected probe should not be greater than 1 / 5 of the label's side length, and the lateral resolution should not be less than 5μm to ensure clear differentiation of the label's boundaries.

[0072] From bottom to top, perform the following steps sequentially for the three labels on the first step and the three labels on the second step:

[0073] ① Label Area Recognition: Utilizing the image or laser scanning function of coordinate measuring software, the software automatically searches for the label's boundaries within a preset measurement area. Since labels and metal substrates typically differ in color or reflectivity, the software can accurately identify the label's complete outline through image binarization or laser reflection intensity thresholding.

[0074] ② Contour point cloud acquisition: Using a planar scanning mode, continuously acquire contour points along the label boundary at a step interval of less than 0.05mm; or acquire surface point clouds by scanning the entire surface of the label using a grid. For tiny labels of 2mm×2mm, the total number of acquisition points should be no less than 100 to ensure fitting accuracy.

[0075] ③ Centroid Fitting: Project the collected boundary point cloud or surface point cloud onto the best-fit plane of the step plane where the label is located, and then use the least squares method to fit the outer rectangle outline of the label. Calculate the geometric center of this rectangle as the centroid coordinates (x, y) of the label. i ,yᵢ,zᵢ), where i=1-6.

[0076] ④ Repeated measurement: To improve reliability, each label is measured three times. Before each measurement, the label boundary is re-identified and points are re-sampled. The average of the three centroid coordinates is taken as the final output.

[0077] The centroid coordinates of the label were obtained by taking multiple measurements and averaging them. The centroid coordinates of the label on the first-level stepped surface are as follows:

[0078] P 11 (X 11 ,Y 11 Z 11 ),P 12 (X 12 ,Y 12 Z 12 ),P 13 (X 13 ,Y 13 Z 13 )

[0079] The centroid coordinates of the second-level stepped surface label are as follows:

[0080] P 21 (X 21 ,Y 21 Z 21 ),P 22 (X 22 ,Y 22 Z 22 ),P 23 (X 23 ,Y 23 Z 23 )

[0081] The coordinates measured by the coordinate measuring machine are ignored due to absolute errors caused by jitter, ambient temperature, etc., and are regarded as the absolute centroid coordinates in the established coordinate system.

[0082] The formula for the distance between two points D1 and D2 in space:

[0083]

[0084] The distance between the centroid coordinates of the tags is calculated for subsequent comparison of coordinate errors after centroid positioning in terahertz scanning. The first number in the letter subscript represents the step number, and the number within the parentheses represents the tag number within the step. For example, D1(1,2) represents the distance between tag 1 and tag 2 on the surface of the first step.

[0085] Calculate the distance between labels on the same stepped surface:

[0086] D1(1,2) D1(2,3) D1(3,1)

[0087] D2(1,2) D2(2,3) D2(3,1)

[0088] like Figure 6The diagram shows the structure of a terahertz imaging system. This system consists of a femtosecond laser, a beam splitter, an optical delay unit, a high-reflectivity mirror, a terahertz probe, a heat-insulating tile sample, a metal stepped block, a skin, a stepper motor, an adapter, and a host machine. The probe contains an off-axis parabolic mirror and a photoconductive antenna. The off-axis parabolic mirror is used to collimate and focus the terahertz beam, while the photoconductive antenna is used to generate and detect reflected terahertz waves. The optical path of this system consists of a femtosecond laser system emitting laser pulses, which are split into a probe beam and a pump beam by a beam splitter. The probe beam, after a certain delay, is received by a photoconductive antenna acting as a detector. The pump beam is guided by a high-reflectivity mirror to the photoconductive antenna acting as a transmitter, radiating terahertz waves. The terahertz pulse signal is then focused onto the surface of a metal step by an off-axis parabolic mirror. After reflection, the terahertz beam is reflected back to the off-axis parabolic mirror at a small angle. The reflected beam then passes through a heat-insulating tile and another off-axis parabolic mirror before being received by the photoconductive antenna and transmitted to a host computer for signal processing. The terahertz probe and a stepper motor are connected via a 3D-printed adapter to secure the probe and motor. A 3D motion control program written in QT drives the terahertz probe to different positions on the tag. The probe emits terahertz waves towards the terahertz tag and the metal step, and receives the time-domain signals reflected from each surface. The host computer displays and records the waveforms. The experimental sample parameters were set as follows: the thickness of the heat insulation tile was 2.5 cm, the label diameter was 2 mm, and a 100 nm thick gold-plated film was deposited on the label surface to enhance the terahertz reflection capability; the distance between the terahertz probe and the metal stepped surface was set to 31.6 cm.

[0089] like Figure 7 The diagram shows a terahertz probe scanning a metal step in the xy two-dimensional plane. The probe scan is a step scan, with each step being 1 mm, and the step distance can be modified in the software. During the scanning process, the probe emits terahertz waves onto the surface of the metal step and detects the reflected terahertz waves. The amplitude of the reflected signal is selected for imaging to obtain the original terahertz image. Subsequent processing and centroid localization are performed on the original image.

[0090] To achieve fast and accurate tag localization, the image processing and localization algorithm flow, such as... Figure 8 As shown, through a series of image preprocessing and feature extraction steps, the sub-pixel-level centroid coordinates of the label are stably and efficiently calculated from the distorted terahertz image. The overall algorithm flow is as follows:

[0091] (1) Image preprocessing: The original terahertz scan image is preprocessed by filtering, interpolation and other operations to suppress system noise and environmental interference and enhance the signal-to-noise ratio of the reflected signal.

[0092] (2) Edge detection and region segmentation: In view of the problem that traditional edge detection algorithms have sparse and broken edges under long focal diffused light spots and cannot be directly used for centroid localization, this algorithm chooses to use a region segmentation-based method to extract the closed region where the label is located, in order to further optimize the traditional gradient-based methods such as Canny, Sobel, and Prewitt. Through binarization, distance transformation and other methods, the closed region of the terahertz label is directly extracted, thereby meeting the needs of fast localization and centroid calculation.

[0093] (3) Centroid localization: Within the successfully segmented closed area, a weighted average algorithm is used to comprehensively consider the gray value distribution of each pixel in the area and accurately calculate the geometric center of the label to achieve sub-pixel level centroid localization.

[0094] like Figure 9 The image shows the terahertz tag coordinates in the XY two-dimensional plane after image processing and centroid localization. The Z-axis coordinates are measured using the peak time delay in the terahertz time-domain waveform, ultimately yielding the tag's three-dimensional coordinates. The tag centroid coordinates measured by the terahertz system are:

[0095] Coordinates for locating the centroid of the first step surface:

[0096] Q 11 (X 11 ,Y 11 Z 11 ),Q 12 (X 12 ,Y 12 Z 12 ),Q 13 (X 13 ,Y 13 Z 13 )

[0097] Coordinates for locating the centroid of the second-step surface:

[0098] Q 21 (X 21 ,Y 21 Z 21 ),Q 22 (X 22 ,Y 22 Z 22 ),Q 23 (X 23 ,Y 23 Z 23 )

[0099] Calculate the distance between labels on the same stepped surface:

[0100] d1(1,2) d1(2,3) d1(3,1)

[0101] d2(1,2) d2(2,3) d2(3,1)

[0102] As shown in Table 1, using the measurement results of the coordinate measuring machine as a benchmark, the positioning accuracy of the terahertz imaging system in the XY two-dimensional plane is calibrated and evaluated by calculating the difference between the distance between the centroids of the two tags measured by the coordinate measuring machine and the terahertz imaging system on the same stepped surface.

[0103] Table 1. Positioning accuracy evaluation of terahertz imaging system in the XY two-dimensional plane

[0104]

[0105] Calculate the distance difference on the z-axis between corresponding points of the first and second steps, as measured by a coordinate measuring machine:

[0106] N1=P 11 (Z 11 )-P 21 (Z 21 N2=P 12 (Z 12 )- P 22 (Z 22 N3=P 13 (Z 13 )-P 23 (Z 23 )

[0107] Calculate the distance difference on the z-axis between the corresponding points of the first and second steps measured using the terahertz imaging system:

[0108] M1=Q 11 (Z 11 )-Q 21 (Z 21 M2=Q 12 (Z 12 )- Q 22 (Z 22 M3=Q 13 (Z 13 )-Q 23 (Z 23 )

[0109] As shown in Table 2, the positioning accuracy of the terahertz imaging system on the z-axis can be calibrated and evaluated by using the difference between the z-axis distance difference measured by the coordinate measuring machine and the z-axis distance difference measured by the terahertz imaging system.

[0110] Table 2. Positioning accuracy evaluation of the terahertz imaging system along the Z-axis

[0111]

[0112] The systems, devices, modules, or units described in the above embodiments can be implemented by computer chips or entities, or by products with certain functions.

[0113] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0114] The above embodiments should be understood as illustrative only and not as limiting the scope of protection of the present invention. After reading the description of the present invention, those skilled in the art can make various alterations or modifications to the present invention, and these equivalent changes and modifications also fall within the scope defined by the claims of the present invention.

Claims

1. A method for calibrating the absolute accuracy of the centroid of a hidden point based on a coordinate measuring machine, characterized in that, Includes the following steps: A calibration carrier with a stepped structure was prepared and terahertz reflective tags were deployed. All labels were measured non-contactly using a coordinate measuring machine to obtain the absolute true coordinates of the three-dimensional centroid of each label. A terahertz reflection imaging system was used to scan and image the calibration carrier, and the three-dimensional centroid coordinates of each label were solved by image processing. Using the true centroid value output by the three coordinate system as a reference, the absolute accuracy calibration of the three-dimensional centroid of the hidden point of the terahertz imaging system is completed by comparing the plane spacing and height difference.

2. The method for calibrating the absolute accuracy of the centroid of a hidden point based on a coordinate measuring machine according to claim 1, characterized in that, The preparation of the calibration carrier with a stepped structure and the deployment of terahertz reflection tags specifically includes: Preparation of calibration carrier and micro terahertz reflective tags: Two-stage aluminum metal stepped blocks are fabricated as calibration carriers. The metal stepped blocks contain two steps with heights of 10 mm and 20 mm, respectively. Each step is 50 mm long and 100 mm wide. A three-layer terahertz reflective tag with a size of 2 mm × 2 mm is prepared. The tag consists of a gold reflective layer, a chromium adhesive layer, and a silicon dioxide substrate layer from top to bottom. Three tags are evenly pasted onto the surface of the first step and three tags are evenly pasted onto the surface of the second step using aviation adhesive.

3. The method for calibrating the absolute accuracy of the centroid of a hidden point based on a coordinate measuring machine according to claim 1, characterized in that, S2, obtaining the true value of the label's three-dimensional absolute centroid based on a coordinate measuring machine, specifically includes: S201. Fix the metal stepped block with the label to the worktable of the coordinate measuring machine. Establish the workpiece coordinate system O-XYZ with the bottom surface of the stepped block as the reference surface and the two mutually perpendicular side surfaces as auxiliary references. Select a non-contact optical probe with a spot diameter of no more than 0.4 mm and a lateral resolution of ≥5 μm. S202. Perform measurements on 6 tags sequentially: Identify the complete outline of the tag by image binarization and laser reflection intensity threshold segmentation, and collect no less than 100 outline / surface point clouds by scanning with a step spacing of ≤0.05mm; project the point cloud onto the step fitting plane, fit the circumscribed rectangle of the tag using the least squares method, and calculate the geometric center to obtain the single centroid coordinates; repeat the measurement 3 times for each tag, and take the average of the three coordinates as the true value Pi of the three-dimensional absolute centroid of the tag, i=1~6; S203. Based on the spatial two-point distance formula, calculate the standard distance D between the centroids of each label within the same step; and calculate the Z-axis coordinate difference N between the corresponding labels of the two steps as the three-dimensional accuracy calibration benchmark data.

4. The method for calibrating the absolute accuracy of the centroid of a hidden point based on a coordinate measuring machine according to claim 1, characterized in that, S3, setting up a terahertz reflective imaging system and collecting tag imaging data, specifically includes: S301. Construct a femtosecond laser-driven reflective terahertz imaging system. The system includes a femtosecond laser, a beam splitter, an optical delay unit, an off-axis parabolic mirror, a photoconductive antenna probe, a stepper motor, and a host computer. A 2.5cm thick heat-insulating tile is laid on the outside of the metal stepped block to simulate the concealed and obstructed working condition. The distance between the probe and the surface of the stepped block is set to 31.6cm. S302: Control the stepper motor to drive the terahertz probe to perform XY plane step scanning on the metal stepped block in 1mm steps, collect the terahertz time-domain reflection signal at each position, extract the reflection amplitude to generate the original terahertz grayscale image; calculate the Z-axis imaging coordinate of each tag based on the peak time delay of the time-domain waveform.

5. The method for calibrating the absolute accuracy of the centroid of a hidden point based on a coordinate measuring machine according to claim 1, characterized in that, The S4, terahertz image subpixel-level centroid solution, specifically includes: S401. Perform filtering and interpolation preprocessing on the original terahertz image to suppress noise; S402. Use a binarization + distance transform joint region segmentation algorithm to extract the complete closed region of the label; S403. The gray-scale weighted average algorithm is used to calculate the sub-pixel level two-dimensional centroid of the segmented closed region. Combined with the time-domain delay Z-axis coordinate, the three-dimensional centroid coordinates Qi of each tag measured by the terahertz system are obtained, i=1~6. S404. Calculate the label imaging distance d within the same step and the Z-axis coordinate difference M between the labels corresponding to the two steps.

6. The method for calibrating the absolute accuracy of the centroid of a hidden point based on a coordinate measuring machine according to claim 1, characterized in that, The S5, three-dimensional centroid absolute accuracy calibration and evaluation, specifically includes: S501, XY plane accuracy: Compare the standard distance D of the three coordinate system within the same step with the terahertz imaging distance d, the difference is the two-dimensional plane positioning error; S502, Z-axis height accuracy: Compare the Z-axis difference N of the three-coordinate reference with the Z-axis difference M of the terahertz imaging. The difference is the positioning error in the depth direction. S503, combining two-dimensional and depth errors, completes the absolute accuracy calibration of the three-dimensional centroid of the hidden marker points of the high and low steps in the terahertz imaging system.

Images