Long distance large depth of field displacement measurement system

Abstract

This application discloses a long-range, large depth-of-field displacement measurement system, comprising: a collaborative calibration device for generating a coordinate transformation matrix to perform spatial coordinate calibration of the measurement system; a spectral coding target device for providing a unique spectral code for each measurement point to achieve sub-pixel-level positioning of the center of a light spot on a spectral coding target; an imaging acquisition device for acquiring image data of the center of the light spot on the spectral coding target; a deformation monitoring device for obtaining deformation data; and a data processing device for extracting the coordinates of the light spot center based on the image data, calculating the change in pixel coordinates of the light spot center based on the initial coordinates of the light spot center, compensating for the change in pixel coordinates based on the deformation data, and calculating the target displacement of the spectral coding target using the compensated change in pixel coordinates. This application provides clear, full-depth-of-field imaging without blurring or blind spots. It exhibits strong anti-interference capabilities, high measurement accuracy, and stable system operation.

Description

Technical Field

[0001] This application relates to the field of data measurement technology, and in particular to a long-distance, large-depth-of-field displacement measurement system. Background Technology

[0002] Currently, the mainstream technologies for long-distance displacement measurement include three categories: total stations, real-time kinematic (RTK) measuring instruments, and machine vision measurement. Total stations require manual operation, are difficult to aim at long distances, and cannot meet the needs of automated real-time monitoring; RTK measuring instruments only achieve centimeter-level accuracy, are expensive, and are susceptible to environmental signal interference; existing machine vision measurement technologies mostly adopt Gaussian distributed active illumination spectral encoded targets and a single-reference platform with six degrees of freedom compensation schemes, relying on centroid algorithms to extract the light spot center, which suffers from limited depth of field, weak resistance to stray light interference, insufficient accuracy of zoom lens focal length error compensation, and single-reference platform susceptibility to environmental disturbances leading to compensation deviations. At the same time, existing technologies use multi-camera regional monitoring or single-camera scanning monitoring modes, which cannot solve the technical pain points of inconsistent image clarity of multispectral encoded targets under large depth of field and incomplete compensation for small deformations of the platform, making it difficult to meet the sub-millimeter-level measurement requirements in scenarios with distances of over 500 meters and large elevation differences.

[0003] In summary, existing long-distance displacement measurement systems suffer from problems such as blurred imaging, low accuracy, and inability to achieve synchronous and rapid measurement of multispectral coded targets. Summary of the Invention

[0004] This disclosure provides a long-distance, large-depth-of-field displacement measurement system to at least solve the above-mentioned technical problems existing in the prior art.

[0005] According to a first aspect of this application, a long-distance, large-depth-of-field displacement measurement system is provided, comprising: A collaborative calibration device includes a collaborative calibration architecture consisting of a main reference point, a first auxiliary reference point, and a second auxiliary reference point; the collaborative calibration architecture is used to generate a coordinate transformation matrix to perform spatial coordinate calibration of the measurement system. The spectral coding target device includes a spectral coding target disposed at the main reference point, the first auxiliary reference point, the second auxiliary reference point, and the area to be measured. The spectral coding target provides a unique spectral code for each point to be measured within the area to be measured, enabling sub-pixel-level positioning of the light spot center on the spectral coding target across the entire depth of field. An imaging acquisition device is used to acquire image data of the center of a light spot on a spectrally encoded target. Deformation monitoring device is used to monitor the deformation of the degrees of freedom of the measuring platform in real time and obtain deformation data; A data processing device is used to extract the center coordinates of the light spot based on the image data, calculate the change in pixel coordinates of the center of the light spot based on the center coordinates of the light spot at the initial time, compensate for the change in pixel coordinates based on the deformation data, and calculate the target displacement of the spectral coded target using the compensated change in pixel coordinates. The collaborative calibration device, the spectral coding target device, the imaging acquisition device, and the deformation monitoring device are all connected to the data processing device.

[0006] In one possible implementation, the main reference point is equipped with: a laser interferometer, a first camera, a main reference spectral coding target, a wireless synchronization trigger, a BeiDou timing antenna, and an edge computing processor; The first auxiliary reference point is equipped with: a second camera, an auxiliary reference spectral coding target, a wireless synchronization trigger, a BeiDou timing antenna, and an edge computing processor; The equipment used to deploy the first auxiliary reference point is the same as that used to deploy the second auxiliary reference point.

[0007] In one possible implementation, the collaborative calibration architecture is used to generate a coordinate transformation matrix to perform spatial coordinate calibration of the measurement system, including: The BeiDou timing antenna is used for time synchronization settings. The first camera acquires the target image of the main reference spectral coded target, and the second camera simultaneously acquires the target image of the auxiliary reference spectral coded target; The edge computing processor extracts the contour edges and corner features of the target image, and establishes a coordinate transformation matrix using the three-dimensional spatial coordinates of the pre-calibrated main reference point, the first auxiliary reference point, and the second auxiliary reference point; During the displacement measurement process, after acquiring a preset number of images, the coordinate transformation matrix is ​​iteratively updated using the incremental least squares method. Real-time monitoring of the fluctuation of the coordinates of the spectral encoded target feature points of each reference point. When the fluctuation of the coordinates of the spectral encoded target feature points of any reference point continuously exceeds the preset fluctuation threshold, the system switches to the dual reference calibration mode and automatically returns to the three reference collaborative mode after the abnormal reference is restored. Among them, the dual-reference calibration mode is a working mode in which the reference with the failed coordinate fluctuation is discarded and the remaining two reference points are used to reconstruct the coordinate transformation matrix when the coordinate fluctuation of a single reference point fails. The three-reference coordination mode is a working mode in which the coordinate transformation matrix is ​​updated using the three reference points when all three reference points are working normally.

[0008] In one embodiment, the spectral coding target includes: a carbon fiber composite substrate and a spectral coding target surface; the surface of the substrate is treated with matte black anodizing. The spectral coding target has an array of light-emitting modules distributed on its surface. The back of the spectral coding target has a built-in driving circuit board, a lithium battery, and a wireless communication module. The driving circuit board is used to drive the array of light-emitting modules to emit light. The array of light-emitting modules consists of a white reference light-emitting module at the center of the array, infrared light-emitting modules at the four corner points of the array edge, and red, green, blue, and infrared light-emitting modules at the remaining 20 points. The array light-emitting module is based on four spectra: red, green, blue, and infrared. Each spectral target is independently encoded using encoding rules, and each spectral code corresponds to a unique spectral combination.

[0009] In one possible implementation, extracting the center coordinates of the light spot based on the image data includes: Target images are acquired based on the master reference synchronous pulse dimming mode; The acquired target image is processed in grayscale and a segmentation threshold is calculated for binarization to obtain a binarized image of the spot region. Traverse all pixels of the image of the binarized region of the light spot and calculate the gray-level centroid coordinates; Using the gray-scale centroid coordinates as the center, an edge detection algorithm is used to extract the sub-pixel edge contour of the light spot; Calculate the geometric center of the sub-pixel edge contour of the light spot to obtain the coordinates of the geometric center; The gray-scale centroid coordinates and the geometric center coordinates are weighted and fused to obtain the center coordinates of the light spot.

[0010] In one possible implementation, the acquisition of target images based on the master reference synchronous pulse dimming mode includes: The wireless synchronization trigger deployed on the main reference outputs a synchronization pulse signal to synchronously control the luminescence state of the array luminescence modules of all spectral encoded targets. Automatically switches between dual lighting modes based on measurement distance and ambient light intensity; When the measured distance is greater than or equal to the preset distance or the light intensity is less than the preset light intensity, the individual spectral coded targets are lit up in the coding order, and the first camera acquires the target images one by one. When the measured distance is less than the preset distance or the light intensity is greater than or equal to the preset light intensity, all spectral-coded targets are lit up simultaneously, and the first camera simultaneously acquires images of all targets.

[0011] In one embodiment, the deformation monitoring device includes: a three-dimensional tilt sensor, multiple triaxial vibration sensors, and a lidar; the real-time monitoring platform measures the degrees of freedom deformation to obtain deformation data, including: The first data was acquired using a three-dimensional tilt sensor; Second data was acquired using multiple triaxial vibration sensors; Acquire third-party data using lidar; The first, second, and third data are sequentially filtered to remove noise, least squares to remove trend terms, and interpolated to align the data to remove outliers. Extended Kalman filtering is applied to the first, second, and third data after removing outliers to fuse the data and output the deformation data of the test platform with six degrees of freedom.

[0012] In one possible implementation, compensating for the change in pixel coordinates based on the deformation data includes: Acquire pixel coordinate changes, deformation data, spatial coordinates of the spectral encoded target, and coordinates of the station center; The compensation-adjusted pixel coordinate change is calculated using the following method.

[0013] in, , Spectral coding target before compensation Pi Changes in the center column and row pixels of the light spot; , These are the compensated spectral encoded targets. Pi Changes in the center column and row pixels of the light spot; X i , Y i , Z i ) as a spectral coding target Pi Three-dimensional spatial coordinates; X c , Y c , Z c (where ) represents the three-dimensional spatial coordinates of the station center); Δ α c Δ β c Δ γ c These represent the deformations of the measuring platform around the X, Y, and Z axes, respectively; Δ X c Δ Z c These represent the translational deformations of the measuring platform along the X and Z axes, respectively.

[0014] In one possible implementation, calculating the target displacement of the spectral-coded target using the compensated pixel coordinate change includes: Based on the compensated pixel coordinate change, real-time focal length, camera pixel size, and distance from the measuring platform to the spectral coding target, the horizontal and vertical displacements of the spectral coding target to be measured are calculated. The horizontal and vertical displacements of the spectral coding target to be measured are calculated using the following method.

[0015] in, Spectral coding target Pi Horizontal displacement; Spectral coding target Pi Vertical displacement; , To compensate for the large changes in pixel coordinates; For the measurement station to the spectral coding target Pi Straight-line distance; pixel The pixel size of the first camera; f This is the real-time focal length of the first camera.

[0016] In one embodiment, the data processing apparatus is further configured to: The rate of change of pixel coordinates is verified, the abrupt change of target displacement is verified, and the correlation of multispectral coded targets is verified.

[0017] The technical solution of this application provides clear, full-view depth imaging without blurring or blind spots. It exhibits strong anti-interference capabilities, high measurement accuracy, and stable system operation.

[0018] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of this application, nor is it intended to limit the scope of this application. Other features of this application will become readily apparent from the following description. Attached Figure Description

[0019] The above and other objects, features, and advantages of exemplary embodiments of this application will become readily apparent from the following detailed description taken in conjunction with the accompanying drawings. Several embodiments of this application are illustrated in the drawings by way of example and not limitation, in which: In the accompanying drawings, the same or corresponding reference numerals indicate the same or corresponding parts.

[0020] Figure 1 A schematic diagram of the structure of the long-distance, large depth-of-field displacement measurement system in an embodiment of this application is shown. Detailed Implementation

[0021] To make the objectives, features, and advantages of this application more apparent and understandable, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0022] In the following description, the terms "first" and "second" are used merely to distinguish similar objects and do not represent a specific ordering of objects. It is understood that "first" and "second" may be interchanged in a specific order or sequence where permitted, so that the embodiments of this application described herein can be implemented in an order other than that illustrated or described herein.

[0023] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0024] The following description, in conjunction with the accompanying drawings, introduces a long-distance, large-depth-of-field displacement measurement system provided in this application.

[0025] like Figure 1 As shown, this application provides a long-distance, large-depth-of-field displacement measurement system, comprising: A collaborative calibration device includes a collaborative calibration architecture consisting of a main reference point, a first auxiliary reference point, and a second auxiliary reference point; the collaborative calibration architecture is used to generate a coordinate transformation matrix to perform spatial coordinate calibration of the measurement system. The spectral coding target device includes a spectral coding target disposed at the main reference point, the first auxiliary reference point, the second auxiliary reference point, and the area to be measured. The spectral coding target provides a unique spectral code for each point to be measured within the area to be measured, enabling sub-pixel-level positioning of the light spot center on the spectral coding target across the entire depth of field. An imaging acquisition device is used to acquire image data of the center of a light spot on a spectrally encoded target. Deformation monitoring device is used to monitor the deformation of the degrees of freedom of the measuring platform in real time and obtain deformation data; A data processing device is used to extract the center coordinates of the light spot based on the image data, calculate the change in pixel coordinates of the center of the light spot based on the center coordinates of the light spot at the initial time, compensate for the change in pixel coordinates based on the deformation data, and calculate the target displacement of the spectral coded target using the compensated change in pixel coordinates. The collaborative calibration device, the spectral coding target device, the imaging acquisition device, and the deformation monitoring device are all connected to the data processing device.

[0026] The long-distance, large depth-of-field displacement measurement system provided in this application consists of a collaborative calibration device, a spectral coding target device, an imaging acquisition device, a deformation monitoring device, and a data processing device. Through low-power LoRa+5G dual-redundant wireless networking, the system achieves millisecond-level data synchronization and interaction. The entire system requires no manual intervention and can stably adapt to complex working conditions such as extreme environments from -40℃ to 85℃, strong electromagnetic interference, and high dust and humidity.

[0027] The collaborative calibration device includes a collaborative calibration architecture consisting of a primary reference point, a first auxiliary reference point, and a second auxiliary reference point. In this application, the system coordinate system adopts a three-dimensional Cartesian right-hand coordinate system, with the primary reference point as the origin O(0,0,0), the horizontal direction to the right as the X-axis, the vertical direction upwards as the Z-axis, and the Y-axis perpendicular to the XZ plane pointing towards the area to be measured as the Y-axis; the coordinates of the rigid marker point at the center of the measuring platform are C(X... c Y c Z c The coordinates of the rigid marker point at the center of each spectral coding target to be tested are Pi(X). i Y i Z i All initial coordinate values ​​are pre-calibrated using a high-precision prism-free total station. Before calibration, the total station is statically calibrated for 24 hours, with ambient temperature fluctuations controlled within ±2℃. The calibration accuracy is strictly controlled to ≤0.5mm, providing an absolutely stable initial spatial reference for subsequent full-process measurement. All coordinate data is stored in the local flash memory of the edge computing processor and simultaneously backed up to the cloud server to prevent data loss.

[0028] This application adopts an equilateral triangular collaborative calibration framework with one main reference point and two auxiliary reference points. All three points are strictly selected on stable bedrock or rigid concrete bases outside the area to be measured. The base depth is ≥1.5m, and the curing period after pouring is ≥28 days to ensure no settlement or shaking. The three points form an equilateral triangle layout, with the side length set to 1 / 2 to 2 / 3 of the straight-line distance from the measuring platform to the main reference point. The three points are fully visible to the measuring platform, with no obstruction or reflection interference, minimizing the impact of environmental disturbances on the stability of the reference. After the reference is set up, it is left to stand for 72 hours until the base is completely stable before subsequent debugging.

[0029] In some embodiments, the main reference point is equipped with: a laser interferometer, a first camera, a main reference spectral coding target, a wireless synchronization trigger, a BeiDou timing antenna, and an edge computing processor; The first auxiliary reference point is equipped with: a second camera, an auxiliary reference spectral coding target, a wireless synchronization trigger, a BeiDou timing antenna, and an edge computing processor; The equipment used to deploy the first auxiliary reference point is the same as that used to deploy the second auxiliary reference point.

[0030] Specifically, the laser interferometer, first camera, main reference spectral encoding target, wireless synchronization trigger, Beidou timing antenna, and edge computing processor are housed in a waterproof and dustproof equipment box. The laser interferometer in this application is a 635nm helium-neon laser interferometer, emitting a 635nm red directional laser with a laser beam divergence angle ≤0.01mrad, a ranging range of 0-1000m, a ranging accuracy ≤0.05mm, and a sampling frequency of 50Hz. It is fixed to the center of the main reference point base by a dedicated rigid bracket with a horizontality ≤0.02mm / m and a verticality ≤0.03mm / m. The laser emission port faces the center of the measuring platform, and the laser beam coincides with the Y-axis, used for real-time measurement of the absolute displacement of the measuring platform in the Y direction. The first camera uses a 12-megapixel (4000×3000) global shutter with a frame rate of 50fps and a pixel size of 3.45μm. It employs a fixed-focus 35mm low-distortion lens with a distortion coefficient ≤0.1%. It is fixed to the laser interferometer via an adjustable gimbal with a horizontal rotation range of ±180° and a pitch range of ±90°. Both horizontal and vertical accuracy are ≤0.02mm / m. The lens optical axis is aligned with the intersection of the spectral coding target of the testing platform and the auxiliary reference spectral coding target, used to acquire images of these targets and achieve cross-calibration between the references. The main reference spectral coding target measures 150mm×150mm, uses a carbon fiber substrate with a thermal expansion coefficient ≤1×10^-6 / ℃, and features a 5×5 array of LED light-emitting units on its surface. It is fixed to the side of the main reference point base using dedicated rigid bolts. The mounting plane is perpendicular to the camera lens optical axis with a verticality ≤0.02mm / m, and the center position is aligned with the camera optical axis with a deviation ≤0.1mm. The wireless synchronization trigger has a triggering accuracy of ≤1μs, outputs a synchronization pulse signal, and synchronizes with the triggers of two auxiliary reference points and the camera trigger on the measuring station. It is fixed inside the equipment box with bolts, and the wiring uses waterproof aviation plugs to prevent moisture-induced short circuits. The Beidou timing antenna has a timing accuracy of ≤1μs and can provide a global time reference for the system. It is installed inside the equipment box with the antenna extending outside the box without obstruction to ensure signal stability. The edge computing processor uses an industrial-grade embedded processor with a main frequency of 1.2GHz, 4GB of memory, and 64GB of storage. It is responsible for main reference data preprocessing, synchronization signal output, and data transmission. It is installed inside a waterproof and dustproof equipment box with an IP67 protection rating and a built-in temperature control module to keep the temperature between 0-50℃.

[0031] The first and second auxiliary reference points are set up in exactly the same way. The second camera is an 8-megapixel (3264×2448) global shutter camera. The frame rate is 30fps, the pixel size is 4.5μm, the ISO sensitivity is 100-102400, the lens is a fixed 50mm lens with a large depth of field and low distortion, the depth of field range is 10-1000m, and the distortion coefficient is ≤0.08%. It is fixed to the center of the auxiliary reference point base by an adjustable angle gimbal. The horizontal and vertical accuracy of the gimbal are both ≤0.02mm / m, and the optical axis of the lens is parallel to the optical axis of the main reference camera with a parallelism of ≤0.01°. The auxiliary reference spectral coding target is 120mm×120mm in size. Its material and array configuration are the same as those of the main reference spectral coding target. The coding is different from that of the main reference spectral coding target. It is fixed to the side of the auxiliary reference point base by rigid bolts. The mounting plane is perpendicular to the optical axis of the camera lens with a perpendicularity of ≤0.02mm / m. The center position is aligned with the optical axis of the camera with a deviation of ≤0.1mm.

[0032] In some embodiments, the collaborative calibration architecture is used to generate a coordinate transformation matrix to perform spatial coordinate calibration of the measurement system, including: The BeiDou timing antenna is used for time synchronization settings. The first camera acquires the target image of the main reference spectral coded target, and the second camera simultaneously acquires the target image of the auxiliary reference spectral coded target; The edge computing processor extracts the contour edges and corner features of the target image, and establishes a coordinate transformation matrix using the three-dimensional spatial coordinates of the pre-calibrated main reference point, the first auxiliary reference point, and the second auxiliary reference point; During the displacement measurement process, after acquiring a preset number of images, the coordinate transformation matrix is ​​iteratively updated using the incremental least squares method. Real-time monitoring of the fluctuation of the coordinates of the spectral encoded target feature points of each reference point. When the fluctuation of the coordinates of the spectral encoded target feature points of any reference point continuously exceeds the preset fluctuation threshold, the system switches to the dual reference calibration mode and automatically returns to the three reference collaborative mode after the abnormal reference is restored. Among them, the dual-reference calibration mode is a working mode in which the reference with the failed coordinate fluctuation is discarded and the remaining two reference points are used to reconstruct the coordinate transformation matrix when the coordinate fluctuation of a single reference point fails. The three-reference coordination mode is a working mode in which the coordinate transformation matrix is ​​updated using the three reference points when all three reference points are working normally.

[0033] Specifically, after the system is powered on, the BeiDou timing antenna completes time synchronization, and all devices enter standby mode. The first camera at the main reference point synchronously outputs a trigger signal, and the second cameras at the two auxiliary reference points synchronously respond, simultaneously acquiring auxiliary reference target images from the two auxiliary reference points, acquiring 10 frames of images at a time. The edge computing processor preprocesses the acquired 10 frames of images, removing blurred, overexposed, and underexposed images, and retaining valid images. Then, an edge feature-corner detection fusion algorithm is used to extract spectral encoded target feature points. First, the Canny edge detection algorithm is used to extract the contour edges of the spectral encoded target, with a high threshold of 200, a low threshold of 100, and a smoothing coefficient of 1.5. Then, the Harris corner detection algorithm is used to extract contour corner points, with a corner response threshold of 0.01 and a non-maximum suppression window of 3×3. Finally, four corner features are extracted from each spectral encoded target, for a total of eight feature points. Based on the pre-calibrated three-dimensional spatial coordinates of the primary reference point O (0, 0, 0), auxiliary reference point A (XA, YA, ZA), and auxiliary reference point B (XB, YB, ZB), a coordinate transformation matrix M between reference points is established. The matrix has a dimension of 4×4 and takes the following form:

[0034] Where R is a 3×3 rotation matrix and T is a 3×1 translation vector, the matrix parameters are solved by the least squares method with a solution error ≤0.001%, the spatial coordinates of the three reference points are unified, and the layout error between references is eliminated; the coordinate transformation matrix M is stored locally on all edge computing processors as the initial calibration parameter. After the initial calibration is completed, the system enters the standby measurement state.

[0035] Then, real-time calibration is performed. During the measurement process, after the first camera acquires 10 frames of images from the measuring platform, it automatically switches to acquire images from two auxiliary reference targets, acquiring 5 frames at a time. The edge computing processor preprocesses the images and extracts feature points, following the same steps as the initial calibration, extracting the coordinates of 8 feature points. The coordinate transformation matrix M is updated in real time, using incremental least squares iterative optimization to reduce computation, with an update time ≤10ms. The two second cameras simultaneously acquire images of the main reference target, extract feature point coordinates, and substitute them into the updated matrix M for reverse verification, calculating the verification error. If the update deviation of matrix M is ≤0.01% and the verification error is ≤0.005%, the calibration is considered valid, and matrix M is updated to the current working parameters. If the update deviation of matrix M is >0.01% or the verification error is >0.005%, the calibration is considered abnormal, triggering a secondary calibration, re-acquiring images, extracting feature points, and calculating the matrix, with the secondary calibration taking ≤20ms, until the calibration is valid.

[0036] This application also sets a threshold for the fluctuation of reference feature point coordinates, specifically ≤0.1mm in the X and Z directions and ≤0.05mm in the Y direction, with a sampling frequency of 100Hz, to monitor the changes in the coordinates of the three reference spectral encoded target feature points in real time. If the coordinates of a single reference feature point exceed the threshold for five consecutive times (0.05s), the reference is determined to be abnormal, and the system immediately locks the abnormal reference and stops receiving its data. The system automatically switches to the dual-reference calibration mode of the remaining two references, recalculates the dual-reference coordinate transformation matrix, and takes ≤15ms to ensure uninterrupted measurement. After the abnormal reference returns to normal (coordinate fluctuation returns to within the threshold, after 10 consecutive samplings), the system automatically returns to the three-reference collaborative mode, updates matrix M, and restores full-precision calibration.

[0037] In some embodiments, the spectral coding target includes: a carbon fiber composite substrate and a spectral coding target surface; the surface of the substrate is treated with matte black anodizing. The spectral coding target has an array of light-emitting modules distributed on its surface. The back of the spectral coding target has a built-in driving circuit board, a lithium battery, and a wireless communication module. The driving circuit board is used to drive the array of light-emitting modules to emit light. The array of light-emitting modules consists of a white reference light-emitting module at the center of the array, infrared light-emitting modules at the four corner points of the array edge, and red, green, blue, and infrared light-emitting modules at the remaining 20 points. The array light-emitting module is based on four spectra: red, green, blue, and infrared. Each spectral target is independently encoded using encoding rules, and each spectral code corresponds to a unique spectral combination.

[0038] This application designs a rectangular array-type multispectral coding target, optimizing the entire process from substrate material, array layout, spectral coding, illumination control, to center extraction, thus solving problems such as poor depth-of-field adaptability, weak resistance to stray light, and low center extraction accuracy. The spectral coding target in this application has a square flat plate structure, with two sizes: 120mm×120mm (auxiliary reference / near-range spectral coding target) and 150mm×150mm (main reference / far-range spectral coding target), with a thickness of 5mm. The substrate is made of high-modulus carbon fiber composite material with a density of 1.6g / cm³, a coefficient of thermal expansion ≤1×10^-6 / ℃, and an elastic modulus ≥150GPa. It is lightweight (≤150g), has good thermal stability, and strong resistance to deformation, avoiding deformation of the spectral coding target caused by temperature changes and wind vibration. The substrate surface is treated with matte black anodizing, with a roughness Ra≤0.8μm, reducing environmental reflection interference.

[0039] In this application, a 5×5 array of light-emitting modules is uniformly distributed on the surface of the spectral-coded target. The array of light-emitting modules uses LED light-emitting units. The module spacing is 20mm. The center module of the array is a white reference light-emitting module, the four corner modules at the edge of the array are infrared light-emitting modules, and the remaining 20 modules are red, green, blue, and infrared LEDs. The unit diameter is 5mm, the emission angle is 15°, the brightness is adjustable from 0-1000cd / m², the response time is ≤1μs, and the consistency error is ≤5%. The back of the spectral-coded target has a built-in driver circuit board, a lithium battery (3.7V / 500mAh), and a wireless communication module. The driver circuit board supports independent control of a single unit. The wireless communication module adopts LoRa, with a communication distance of ≥2km, power consumption of ≤10mW, and a lithium battery life of ≥1 year. The back is waterproof and sealed, with an IP68 protection rating, making it suitable for harsh outdoor environments.

[0040] The LED light-emitting unit uses four spectra (red, green, blue, and infrared). Each spectral target is independently encoded using a 4-bit binary unique coding rule. Each code corresponds to a unique spectral combination, with a code length of 4 bits, enabling 16 different codes. This meets the requirement for simultaneous identification of 16 spectral targets in a single area. The coding rule is as follows: Encoding bit definition: D3D2D1D0, D3=infrared, D2=red, D1=green, D0=blue, 1=light emission, 0=no emission; For example, 0001 (blue light only), 0010 (green light only), 0100 (red light only), 1000 (infrared light only), 0011 (green + blue light), 0101 (red + blue light), 0110 (red + green light), 1001 (infrared + blue light), and so on, with 16 unique and conflict-free codes. Reference unit coding: The central white reference unit is fixed to full-spectrum emission (red + green + blue + infrared), coded as 1111, serving as a universal centering reference for all spectral-coded targets; Corner unit coding: The four corner units at the edges are fixed to infrared emission, coded as 1000, used for rapid identification of the spectral-coded target outline, distinguishing stray light sources in the field (such as lamps, reflective objects, and natural light spots), and improving anti-stray light interference capability by more than 2 times compared to traditional Gaussian spectral-coded targets.

[0041] In some embodiments, extracting the center coordinates of the light spot based on the image data includes: Target images are acquired based on the master reference synchronous pulse dimming mode; The acquired target image is processed in grayscale and a segmentation threshold is calculated for binarization to obtain a binarized image of the spot region. Traverse all pixels of the image of the binarized region of the light spot and calculate the gray-level centroid coordinates; Using the gray-scale centroid coordinates as the center, an edge detection algorithm is used to extract the sub-pixel edge contour of the light spot; Calculate the geometric center of the sub-pixel edge contour of the light spot to obtain the coordinates of the geometric center; The gray-scale centroid coordinates and the geometric center coordinates are weighted and fused to obtain the center coordinates of the light spot.

[0042] The acquisition of target images based on the master reference synchronous pulse dimming mode includes: The wireless synchronization trigger deployed on the main reference outputs a synchronization pulse signal to synchronously control the luminescence state of the array luminescence modules of all spectral encoded targets. Automatically switches between dual lighting modes based on measurement distance and ambient light intensity; When the measured distance is greater than or equal to the preset distance or the light intensity is less than the preset light intensity, the individual spectral coded targets are lit up in the coding order, and the first camera acquires the target images one by one. When the measured distance is less than the preset distance or the light intensity is greater than or equal to the preset light intensity, all spectral-coded targets are lit up simultaneously, and the first camera simultaneously acquires images of all targets.

[0043] The specific parameters of the synchronization pulse signal in this application are: pulse width 100μs, period 10ms, duty cycle 1:100, high level 5V, low level 0V, trigger accuracy ≤1μs, signal transmission delay ≤50μs, ensuring that the emission time of all spectral encoded targets is completely aligned with the imaging time of the first camera, reducing motion blur; the spectral encoded target LEDs adopt pulsed emission, not continuous illumination, which greatly reduces power consumption, with a static power consumption of ≤5mW for a single spectral encoded target and a lithium battery life of ≥1 year.

[0044] This application employs a dual illumination mode switching mechanism. Specifically, the scanning mode and synchronization mode are automatically switched based on the measurement distance and ambient light intensity, with a switching threshold of 500m (distance from the measuring station to the spectral encoded target) and an ambient light intensity threshold of 500 lux. For example, in scanning mode (distance > 500m or light intensity < 500 lux): individual spectral encoded targets are illuminated sequentially according to the encoding order (0001→0010→…→1111), illuminating one spectral encoded target at a time for one pulse cycle (10ms). The first camera sequentially acquires images of individual targets, with a 10ms interval between acquisitions of adjacent spectral encoded targets. In this mode, the signal-to-noise ratio of the spectral encoded target imaging is ≥40dB, effectively avoiding overlapping light spots and stray light interference from distant multispectral encoded targets, thus improving imaging clarity. Synchronous mode (distance ≤ 500m or light intensity ≥ 500 lux) specifically involves all spectral-coded targets illuminating simultaneously with completely synchronized emission timing for one pulse cycle. The first camera acquires images of multiple targets at once, covering 16 spectral-coded targets in a single acquisition, with a measurement frame rate ≥ 50fps, significantly improving the efficiency of close-range measurements and meeting the needs of rapid deformation monitoring.

[0045] This application adopts a master reference synchronous pulse dimming mode, in which a high-precision synchronous pulse signal is output by the master reference point wireless synchronous trigger to synchronously control the LED light emission status of all the spectral coding targets under test, the master reference spectral coding target, and the auxiliary reference spectral coding target, thereby achieving synchronous illumination of the entire system's spectral coding targets and avoiding imaging blur caused by timing misalignment.

[0046] This application also features adaptive brightness adjustment. It monitors the target image brightness in real time and dynamically adjusts the LED brightness to ensure uniform target image brightness without overexposure or underexposure under full depth and intensity conditions. The specific steps are as follows: The first camera acquires the target image, processes it into grayscale, and calculates the mean grayscale value G and the grayscale variance σ, with a grayscale range of 0-255. Brightness thresholds are set: low threshold G1=50, high threshold G2=200, and variance threshold σ0=20. If G < G1 (underexposure): brightness adjustment step size +10%, maximum brightness 100% per cycle, adjusted once every 10ms pulse period until G ≥ G1. If G > G2 (overexposure): brightness adjustment step size -10%, minimum brightness 10% per cycle, adjusted once every 1 pulse period until G ≤ G2. If G1 ≤ G ≤ G2 and σ ≤ σ0 (uniform brightness): maintain the current brightness. If G1≤G≤G2 but σ>σ0 (uneven brightness): fine-tune the brightness by ±5%, adjusting once every 10ms, until σ≤σ0. The brightness adjustment is executed automatically in a closed loop throughout the process, with a response time ≤20ms, ensuring stable imaging quality of the spectral encoded target.

[0047] In this application, the method for extracting the center coordinates of a light spot based on the image data abandons the traditional single centroid algorithm and adopts a gray-level centroid-subpixel edge fusion algorithm to extract the center of the light spot. The extraction accuracy is better than 0.05 pixels, which is twice that of the traditional centroid algorithm. The specific steps are as follows: The first camera acquires a color image of the spectrally coded target, which is then converted to grayscale. The specific grayscale formula is: G = 0.299R + 0.587G + 0.114B, where R, G, and B are the pixel values ​​of the three channels of the color image. The grayscale image histogram is calculated, and the number of pixels at grayscale levels 0-255 is counted to determine the dynamic segmentation threshold T, T = 0.3 × Gmax, where Gmax is the maximum grayscale value in the image, effectively distinguishing the spectrally coded target spot from the background. Then, binarization is performed: pixels with grayscale values ​​≥ T are set to 1 (spot area), and pixels with grayscale values ​​< T are set to 0 (background area). Morphological filtering is then performed: opening operations (erosion followed by dilation) using a 3×3 structuring element are performed to eliminate isolated noise in the background; followed by closing operations (dilation followed by erosion) are performed to fill the voids inside the spot, resulting in a complete and clean binarized spot area.

[0048] Then, coarse extraction of grayscale centroid (rapidly locating the center of the light spot) is performed, specifically as follows: Within the binarized spot area, traversing all pixels, the gray-level centroid coordinates are calculated using the following method: u 0, v 0),

[0049]

[0050] in: N This represents the total number of effective pixels within the light spot area. k For pixel number; u k ,v k ) is the first k The column and row coordinates of a pixel in the image coordinate system, in pixels; I k For the first k The grayscale value of each pixel, ranging from 0 to 255, is calculated in ≤5ms, quickly obtaining the coarse positioning coordinates of the spot center.

[0051] Then, sub-pixel edge fine-tuning is performed to improve edge accuracy. Specifically, this involves using the grayscale center of gravity (…) u 0, v Centered on 0), a 3×3 pixel neighborhood is selected as the fine-localization region, and the Zernike moment subpixel edge detection algorithm is used to extract the subpixel edges of the spot. For example, the Zernike moments of the pixels within the 3×3 neighborhood are calculated, with moment order n=4, angular frequency m=2, and template size 3×3. Edge parameters are solved: edge gradient, edge direction, and subpixel edge coordinates, with subpixel accuracy ≤0.01 pixels. The complete subpixel edge contour of the spot is extracted, obtaining ≥20 edge points. The geometric center of the edge contour is calculated using the following method ( u 1, v1),

[0052]

[0053] in: M The total number of sub-pixel edge points, ( u j , v j ) is the first j Subpixel coordinates of edge points, in pixels.

[0054] Grayscale center ( u 0, v 0) and subpixel edge center ( u 1, v 1) Weighted fusion is performed, and the weight coefficients are calibrated through a large amount of measured data: gray-scale centroid weight ω0=0.4, sub-pixel edge center weight ω1=0.6, and the precise center coordinates (u,v) of the spot are finally calculated in the following way. u =0.4 u 0+0.6 u 1, v =0.4 v 0+0.6 v 1.

[0055] After fusion, the accuracy of the center coordinates is ≤0.05 pixels, the calculation time is ≤8ms, and there is no manual intervention throughout the process. It automatically outputs high-precision center coordinates for subsequent displacement calculation.

[0056] This application adopts a multi-channel imaging architecture with one first camera and two second cameras to replace the traditional single-camera scanning or multi-camera regional mode. The first camera and the second camera work together to cover an ultra-large depth of field of 10-1000m, solving the problems of inconsistent image clarity of spectral encoded targets under large depth of field and low signal-to-noise ratio of long-distance imaging.

[0057] This application also preprocesses the acquired images. First, dark current correction is performed. Specifically, all cameras acquire 10 frames of completely black images (lens cap closed, no light); the dark current mean matrix D of the completely black images is calculated, with the matrix size matching the image pixel size, and each pixel position corresponding to the dark current mean; for each subsequently acquired frame of the original image Iraw, the dark current mean matrix D is subtracted pixel by pixel to obtain the corrected image I1 = Iraw - D; dark current correction is automatically recalibrated monthly, and temporary calibration is automatically triggered when the ambient temperature changes by ≥5℃ to ensure correction accuracy. Furthermore, the first camera undergoes lens distortion correction before image acquisition to eliminate lens geometric distortion. Specifically, all cameras undergo distortion calibration before leaving the factory, obtaining the intrinsic parameter matrix K and distortion coefficients (…).k 1, k 2, p 1, p 2) K is a 3×3 matrix. k 1. k 2 represents the radial distortion coefficient. p 1. p 2 represents the tangential distortion coefficient; then, a polynomial distortion correction model is used to correct the distortion of the dark current-corrected image I1. The model formula is:

[0058] Where: (x,y) are the ideal, distortion-free pixel coordinates, ( x dist , y dist ) represents the distorted pixel coordinates, r² = x² + y²; the pixels are remapped using a bilinear interpolation algorithm to generate the distorted image I2, with a distortion error of ≤0.01 pixels; the distortion correction parameters are permanently stored, requiring no repeated calibration, and are stable and effective in the long term.

[0059] The distortion-corrected image I2 was divided into 8×8 pixel sub-blocks with a 50% overlap to avoid block artifacts. A gray-level histogram was calculated for each sub-block, with 256 bins and a cropping threshold of 40 (to limit contrast and prevent noise amplification). Histogram equalization was performed on each sub-block to map gray values ​​and enhance local contrast. Bilinear interpolation was used to fuse the results of adjacent sub-blocks, generating a gray-level equalized image I3 to address uneven brightness and local overexposure / over-brightness issues in images with large depth of field. Bilateral filtering was then performed on the gray-level equalized image I3, with a filter kernel size of 5×5. The spatial domain standard deviation σs=2 to control the spatial neighborhood range; the gray-level domain standard deviation σr=10 to control gray-level similarity. Bilateral filtering was performed using the following method.

[0060] Weight w = ws wr and ws are spatial weights, and wr is grayscale weight; The filtered image I4 retains the edge details of the spectral encoded target while eliminating Gaussian noise and salt-and-pepper noise, with a noise suppression rate of ≥95%. Then, spectral separation is performed.

[0061] Specifically, in this application, the first camera: based on the spectral encoding spectrum of the spectral encoded target, uses a corresponding bandpass filtering algorithm to separate the spectral image. For example, when the spectral encoded target is in the red spectrum, pixels in the 620-630nm band are retained, and stray light in other bands is filtered out to generate the spectrally separated image I5. The second camera: uses an infrared cutoff filtering algorithm to filter the infrared band above 700nm, retaining the visible light band of 400-700nm, to generate the spectrally separated image I5. The signal-to-noise ratio of the spectrally separated image is ≥40dB, the spectral encoded target spot is clear, the background is clean, and there is no stray light interference, which can be directly used for spectrally encoded target center extraction.

[0062] In some embodiments, the deformation monitoring device includes: a three-dimensional tilt sensor, multiple triaxial vibration sensors, and a lidar; the real-time monitoring platform measures the degrees of freedom deformation to obtain deformation data, including: The first data was acquired using a three-dimensional tilt sensor; Second data was acquired using multiple triaxial vibration sensors; Acquire third-party data using lidar; The first, second, and third data are sequentially filtered to remove noise, least squares to remove trend terms, and interpolated to align the data to remove outliers. Extended Kalman filtering is applied to the first, second, and third data after removing outliers to fuse the data and output the deformation data of the test platform with six degrees of freedom.

[0063] This application abandons the traditional six-degree-of-freedom monitoring mode of two-dimensional inclinometer + laser displacement measuring instrument, and adopts a multi-dimensional deformation monitoring method of three-dimensional inclinometer + triaxial vibration sensor + lidar, which fully covers the three-dimensional translation (ΔXc, ΔYc, ΔZc) and three-dimensional rotation (Δαc, Δβc, Δγc) of the measuring platform. The monitoring accuracy is improved by more than 30% compared with the traditional solution. After data fusion, the translation accuracy is ≤0.05mm and the rotation accuracy is ≤0.3 arcseconds, completely eliminating the influence of the small deformation of the measuring platform on the measurement results.

[0064] The three-dimensional tilt sensor in this application has a measurement range of ±5°, an angular resolution of 0.1 arcseconds, a measurement accuracy of ≤0.5 arcseconds, a sampling frequency of 100Hz, an output signal of RS485, and a communication rate of 9600bps. It is rigidly fixed to the center plane of the measuring platform, seamlessly fitting the platform, with a mounting plane levelness ≤0.02mm / m. The sensor's X, Y, and Z axes are strictly aligned with the system coordinate system, with an alignment deviation ≤0.01°. The three-dimensional tilt sensor acquires the platform's tilt angles Δαc around the X-axis, Δβc around the Y-axis, and Δγc around the Z-axis in real time, and transmits the data to the edge computing processor in real time.

[0065] This application uses four triaxial vibration sensors, symmetrically arranged at the four corners of the measuring platform to form a distributed monitoring network. The measurement range is ±2g, the acceleration resolution is 1μg, the measurement accuracy is ≤0.001g, the sampling frequency is 200Hz, the output signal is RS485, and the communication rate is 19200bps. The triaxial vibration sensors are rigidly fixed to the bottom of the four corners of the measuring platform, rigidly connected to the platform, with a mounting plane perpendicularity ≤0.02mm / m. The X, Y, and Z axes of the sensors are aligned with the system coordinate system with a deviation ≤0.01°. The triaxial vibration sensors acquire X, Y, and Z axis acceleration data in real time, calculate the vibration displacement using a quadratic integration algorithm with an integration time constant of 0.1s and a filter cutoff frequency of 10Hz to eliminate high-frequency vibration interference, outputting the three-dimensional vibration displacement at the four corners of the measuring platform. The average value is used to obtain the initial values ​​ΔXc0, ΔYc0, and ΔZc0 of the three-dimensional translation at the center of the measuring platform.

[0066] The lidar in this application emits at a wavelength of 905 nm, with a horizontal scanning range of 360°, a vertical scanning range of ±15°, an angular resolution of 0.1°, a ranging range of 0-200 m, a ranging accuracy of ≤0.1 mm, a sampling frequency of 50 Hz, and outputs 3D point cloud data. The lidar is rigidly fixed to the top of the main reference point, with a levelness of ≤0.02 mm / m. Its scanning direction is directly facing the measuring platform, without obstruction, and it scans the platform surface in real time to generate a 3D point cloud model of the platform. The lidar employs an Iterative Closest Point (ICP) point cloud registration algorithm to register the real-time point cloud with the initial reference point cloud, calculating the overall 3D translations ΔXc1, ΔYc1, ΔZc1 and 3D rotations Δαc1, Δβc1, Δγc1 of the measuring platform, which serve as calibration benchmarks for the translation and rotation data.

[0067] This application employs a moving average filter with a window size of 5 on the raw data from each sensor to eliminate high-frequency random noise; for data detrending, least squares linear fitting is used to eliminate trend errors caused by long-term sensor drift; for data time alignment, sensor data with different sampling frequencies are interpolated and aligned to 100Hz based on BeiDou time synchronization, with a time error ≤1μs; for abnormal data removal, a fluctuation threshold is set for each sensor data, and data exceeding the threshold is judged as abnormal and replaced by interpolation of adjacent valid data.

[0068] Using the six-degree-of-freedom deformation of the measuring platform (ΔXc, ΔYc, ΔZc, Δαc, Δβc, Δγc) as state variables, a six-dimensional linear state equation is established to describe the deformation variation over time: X(k) = A·X(k) 1)+W(k 1) in: X(k)=[ΔXc(k),ΔYc(k),ΔZc(k),Δαc(k),Δβc(k),Δγc(k)]^T, where k is the sampling time; A is a 6×6 state transition matrix, calibrated by the deformation dynamics characteristics of the test platform, A=diag(1,1,1,1,1,1); W(k-1) is the process noise, zero-mean Gaussian white noise, with a covariance matrix Q=diag(1e-8,1e-8,1e-8,1e-12,1e-12,1e-12).

[0069] Using the measurement data from each sensor as the observed variable, a 6-dimensional observation equation is established to establish the mapping relationship between the state variable and the observed variable: Z(k) = H·X(k) + V(k) Where: Z(k)=[ΔXc0(k),ΔYc0(k),ΔZc0(k),Δαc(k),Δβc(k),Δγc(k)]^T, which includes the initial translation value of the triaxial vibration sensor and the rotation value of the three-dimensional tilt sensor; H is a 6×6 observation matrix, H=diag(1,1,1,1,1,1); V(k) is the observation noise, zero-mean Gaussian white noise, and the covariance matrix R=diag(1e-6,1e-6,1e-6,1e-10,1e-10,1e-10).

[0070] State prediction is performed using the following method:

[0071] Covariance prediction is performed using the following method:

[0072] Kalman gain is performed in the following way: K(k) = P(k∣k) 1)·HT·(H·P(k∣k 1)·HT+R) 1 The status is updated in the following way:

[0073] The covariance is updated using the following method: P(k∣k)=(I K(k)·H)·P(k∣k 1) The overall deformation data obtained by lidar registration is used to perform secondary calibration on the filtering results to correct the deviation. Finally, the optimal estimated values ​​of the fused six-degree-of-freedom deformation are output: ΔXc, ΔYc, ΔZc, Δαc, Δβc, Δγc. The data accuracy is: translation ≤ 0.05mm and rotation ≤ 0.3 arcseconds.

[0074] This application employs an extended Kalman filter (EKF) fusion algorithm to fuse multi-source deformation data collected by a three-dimensional tilt sensor, a three-axis vibration sensor, and a lidar, eliminating single-sensor errors, drift, and interference, and outputting high-precision, high-stability six-degree-of-freedom deformation data.

[0075] This application abandons the traditional fixed-spacing spot-based focal length deduction method and adopts a focal length deduction algorithm with spatial constraints from three or more known coordinate spectral encoding targets. This algorithm deduces the real-time focal length f of the zoom lens of the first camera in real time, eliminating zoom lens focal length drift and zoom error. The deduction accuracy is ≤0.001mm, as detailed below: Three uniformly distributed, unobstructed, and clearly imaged spectral coding targets Pi(Xi,Yi,Zi) (i=1,2,3) were selected. Their three-dimensional spatial coordinates were pre-calibrated using a total station with an accuracy of ≤0.5mm. The distance difference between the measuring station and each spectral coding target was ≥50m to ensure the effectiveness of the spatial constraints.

[0076] The first camera simultaneously acquires images of three targets. After preprocessing and center extraction, the center pixel coordinates (ui,vi) of each spot are obtained (i=1,2,3). The current center coordinates C(Xc,Yc,Zc) of the measuring station and the pixel size of the first camera are read. pixel (μm level, fixed value, factory calibration), the straight-line distance Si from the measuring platform to each spectral encoded target (measured in real time by lidar, with an accuracy of ≤0.1mm).

[0077] Based on the principle of pinhole imaging, a mapping equation between focal length f and pixel coordinates and spatial coordinates is established:

[0078]

[0079] i=1,2,3 in, u i 、v i These represent the column and row pixel coordinates of the center of the spectral encoded target Pi spot; X i 、Z i Spectral coding targets Pi The X and Z axis spatial coordinates; X c 、Z c These are the spatial coordinates of the X and Z axes of the measuring station center, respectively. f This is the real-time focal length of the first camera. S i The distance is the straight-line distance from the measuring platform to the spectral coding target Pi. pixelThis refers to the pixel size of the first camera.

[0080] Then, the optimal focal length is solved using the least squares method. Specifically, the six equations for the three spectral-coded targets are solved simultaneously, and the objective function is constructed using the least squares method to solve for the optimal focal length. f : The objective function is

[0081] right f Taking the derivative and setting it to zero, we obtain the normal equation:

[0082] Iterative solution: The Gauss-Newton iterative method is used to solve the problem. The initial iteration value is the nominal focal length of the lens. The convergence condition of the iteration is: the difference in focal length between two iterations ≤ 0.001mm, the number of iterations ≤ 10, and the final output is the real-time focal length f. The focal length parameter in the displacement calculation formula is updated to eliminate zoom error.

[0083] A six-degree-of-freedom deformation full compensation model is established. The fused six-degree-of-freedom deformation data of the measuring station (ΔXc, ΔYc, ΔZc, Δαc, Δβc, Δγc) are substituted into the model to accurately compensate for the influence of three-dimensional translation and rotation of the measuring station on the changes in spectral encoded target pixels, eliminating systematic errors caused by measuring station deformation. The model formula derivation is based on spatial coordinate transformation and fully covers the six-degree-of-freedom deformation. The compensation-adjusted pixel coordinate change is calculated using the following method.

[0084] in, , Spectral coding target before compensation Pi Changes in the center column and row pixels of the light spot; , These are the compensated spectral encoded targets. Pi Changes in the center column and row pixels of the light spot; X i , Y i , Z i ) as a spectral coding target Pi Three-dimensional spatial coordinates; X c , Y c , Z c (where ) represents the three-dimensional spatial coordinates of the station center); Δ α c Δ β c Δ γc These represent the deformations of the measuring platform around the X, Y, and Z axes, respectively; Δ X c Δ Z c These represent the translational deformations of the measuring platform along the X and Z axes, respectively.

[0085] The compensation process is as follows: ① Real-time reading of spectral encoded target pixel change values , ② Read the six-degree-of-freedom deformation data of the measuring platform and convert the angle values ​​to radians; ③ Substitute the values ​​into the compensation formula to calculate the pixel change value after compensation. , ④ The compensated data is directly used for subsequent displacement calculations, with a compensation time of ≤5ms and is executed in real time.

[0086] Based on the compensated pixel change value, real-time focal length, pixel size of the first camera, and distance from the measuring platform to the spectral coding target, the horizontal displacement Δ of the spectral coding target Pi to be measured is calculated. X i (X-axis) and vertical displacement Δ Z i (Z-axis), calculation accuracy: horizontal ≤ 0.1mm, vertical ≤ 0.08mm, the steps are as follows: The horizontal and vertical displacements of the spectral coding target to be measured are calculated using the following method.

[0087] in, Spectral coding target Pi Horizontal displacement; Spectral coding target Pi Vertical displacement; , To compensate for the large changes in pixel coordinates; For the measurement station to the spectral coding target Pi Straight-line distance; pixel The pixel size of the first camera; f This is the real-time focal length of the first camera.

[0088] In some embodiments, the data processing device is also used for The rate of change of pixel coordinates is verified, the abrupt change of target displacement is verified, and the correlation of multispectral coded targets is verified.

[0089] This application employs a three-level data verification mechanism to verify pixel changes, abrupt shifts, and multispectral coded target correlation throughout the entire process, eliminating abnormal data and ensuring the reliability of measurement data. The verification time is ≤8ms / frame. Level 1 verification: Pixel change rate verification; The rate of change of spectral encoded target pixels is ≤5 pixels / second; Calculate the pixel change rate of adjacent frames. If the rate is greater than 5 pixels / second, it is judged as abnormal, and the data of that frame is directly discarded and replaced with the valid data of the previous frame.

[0090] Secondary verification: displacement abrupt change verification; Horizontal displacement abrupt change ≤ 0.5 mm / frame, vertical displacement abrupt change ≤ 0.3 mm / frame; Calculate the displacement change between adjacent frames. If a sudden change exceeds the threshold, it is considered abnormal, triggering a second measurement. The image is re-acquired, the center is extracted, and the displacement is calculated. If the second calculation is still abnormal, the data is discarded.

[0091] Level 3 verification: Multispectral coded target correlation verification; The displacement trends of multiple spectral coded targets within the same monitoring area (within a 5m radius) should be consistent. A displacement difference of more than 200% between a single spectral coded target and its surrounding targets is considered abnormal. Processing: Abnormal data was removed and replaced with the mean displacement of the three surrounding valid spectral encoded targets to ensure data continuity.

[0092] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage and optical storage) containing computer-usable program code.

[0093] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0094] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to operate in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction methods implemented in a process. Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0095] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0096] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A long-distance, large depth-of-field displacement measurement system, characterized in that, include: The collaborative calibration device includes a collaborative calibration architecture consisting of a main reference point, a first auxiliary reference point, and a second auxiliary reference point. The collaborative calibration architecture is used to generate a coordinate transformation matrix to perform spatial coordinate calibration of the measurement system; The spectral coding target device includes a spectral coding target disposed at the main reference point, the first auxiliary reference point, the second auxiliary reference point, and the area to be measured. The spectral coding target provides a unique spectral code for each point to be measured within the area to be measured, enabling sub-pixel-level positioning of the light spot center on the spectral coding target across the entire depth of field. An imaging acquisition device is used to acquire image data of the center of a light spot on a spectrally encoded target. Deformation monitoring device is used to monitor the deformation of the degrees of freedom of the measuring platform in real time and obtain deformation data; A data processing device is used to extract the center coordinates of the light spot based on the image data, calculate the change in pixel coordinates of the center of the light spot based on the center coordinates of the light spot at the initial time, compensate for the change in pixel coordinates based on the deformation data, and calculate the target displacement of the spectral coded target using the compensated change in pixel coordinates. The collaborative calibration device, the spectral coding target device, the imaging acquisition device, and the deformation monitoring device are all connected to the data processing device.

2. The long-distance, large depth-of-field displacement measurement system according to claim 1, characterized in that, The main reference point is equipped with: a laser interferometer, a first camera, a main reference spectral coding target, a wireless synchronization trigger, a BeiDou timing antenna, and an edge computing processor; The first auxiliary reference point is equipped with: a second camera, an auxiliary reference spectral coding target, a wireless synchronization trigger, a BeiDou timing antenna, and an edge computing processor; The equipment used to deploy the first auxiliary reference point is the same as that used to deploy the second auxiliary reference point.

3. The long-distance, large depth-of-field displacement measurement system according to claim 2, characterized in that, The collaborative calibration architecture is used to generate a coordinate transformation matrix to perform spatial coordinate calibration of the measurement system, including: The BeiDou timing antenna is set to synchronize time. The first camera acquires the target image of the main reference spectral coded target, and the second camera simultaneously acquires the target image of the auxiliary reference spectral coded target; The edge computing processor extracts the contour edges and corner features of the target image, and establishes a coordinate transformation matrix using the three-dimensional spatial coordinates of the pre-calibrated main reference point, the first auxiliary reference point, and the second auxiliary reference point; During the displacement measurement process, after acquiring a preset number of images, the coordinate transformation matrix is ​​iteratively updated using the incremental least squares method. Real-time monitoring of the fluctuation of the coordinates of the spectral encoded target feature points of each reference point. When the fluctuation of the coordinates of the spectral encoded target feature points of any reference point continuously exceeds the preset fluctuation threshold, the system switches to the dual reference calibration mode and automatically returns to the three reference collaborative mode after the abnormal reference is restored. Among them, the dual-reference calibration mode is a working mode in which the reference with the failed coordinate fluctuation is discarded and the remaining two reference points are used to reconstruct the coordinate transformation matrix when the coordinate fluctuation of a single reference point fails. The three-reference coordination mode is a working mode in which the coordinate transformation matrix is ​​updated using the three reference points when all three reference points are working normally.

4. The long-distance, large depth-of-field displacement measurement system according to claim 2, characterized in that, The spectral coding target includes: a carbon fiber composite substrate and a spectral coding target surface; the surface of the substrate is treated with matte black anodizing. The spectral coding target has an array of light-emitting modules distributed on its surface. The back of the spectral coding target has a built-in driving circuit board, a lithium battery, and a wireless communication module. The driving circuit board is used to drive the array of light-emitting modules to emit light. The array of light-emitting modules consists of a white reference light-emitting module at the center of the array, infrared light-emitting modules at the four corner points of the array edge, and red, green, blue, and infrared light-emitting modules at the remaining 20 points. The array light-emitting module is based on four spectra: red, green, blue, and infrared. Each spectral target is independently encoded using encoding rules, and each spectral code corresponds to a unique spectral combination.

5. The long-distance, large depth-of-field displacement measurement system according to claim 4, characterized in that, The step of extracting the center coordinates of the light spot based on the image data includes: Target images are acquired based on the master reference synchronous pulse dimming mode; The acquired target image is processed in grayscale and a segmentation threshold is calculated for binarization to obtain a binarized image of the spot region. Traverse all pixels of the image of the binarized region of the light spot and calculate the gray-level centroid coordinates; Using the gray-scale centroid coordinates as the center, an edge detection algorithm is used to extract the sub-pixel edge contour of the light spot; Calculate the geometric center of the sub-pixel edge contour of the light spot to obtain the coordinates of the geometric center; The gray-scale centroid coordinates and the geometric center coordinates are weighted and fused to obtain the center coordinates of the light spot.

6. The long-distance, large depth-of-field displacement measurement system according to claim 5, characterized in that, The acquisition of target images based on the master reference synchronous pulse dimming mode includes: The wireless synchronization trigger deployed on the main reference outputs a synchronization pulse signal to synchronously control the luminescence state of the array luminescence modules of all spectral encoded targets. Automatically switches between dual lighting modes based on measurement distance and ambient light intensity; When the measured distance is greater than or equal to the preset distance or the light intensity is less than the preset light intensity, the individual spectral coded targets are lit up in the coding order, and the first camera acquires the target images one by one. When the measured distance is less than the preset distance or the light intensity is greater than or equal to the preset light intensity, all spectral-coded targets are lit up simultaneously, and the first camera simultaneously acquires images of all targets.

7. The long-distance, large depth-of-field displacement measurement system according to claim 1, characterized in that, The deformation monitoring device includes: a three-dimensional tilt sensor, multiple triaxial vibration sensors, and a lidar; the real-time monitoring platform measures the degrees of freedom deformation to obtain deformation data, including: The first data was acquired using a three-dimensional tilt sensor; Second data was acquired using multiple triaxial vibration sensors; Acquire third-party data using lidar; The first, second, and third data are sequentially filtered to remove noise, least squares to remove trend terms, and interpolated to align the data to remove outliers. Extended Kalman filtering is applied to the first, second, and third data after removing outliers to fuse the data and output the deformation data of the test platform with six degrees of freedom.

8. The long-distance, large depth-of-field displacement measurement system according to claim 1, characterized in that, The compensation for the change in pixel coordinates based on the deformation data includes: Acquire pixel coordinate changes, deformation data, spatial coordinates of the spectral encoded target, and coordinates of the station center; The compensation-adjusted pixel coordinate change is calculated using the following method. in, , Spectral coding target before compensation Pi Changes in the center column and row pixels of the light spot; , These are the compensated spectral encoded targets. Pi Changes in the center column and row pixels of the light spot; X i , Y i , Z i ) as a spectral coding target Pi Three-dimensional spatial coordinates; X c , Y c , Z c ) represents the three-dimensional spatial coordinates of the center of the measuring platform; Δ α c Δ β c Δ γ c These represent the deformations of the measuring platform around the X, Y, and Z axes, respectively; Δ X c Δ Z c These represent the translational deformations of the measuring platform along the X and Z axes, respectively.

9. The long-distance, large depth-of-field displacement measurement system according to claim 1, characterized in that, The calculation of the target displacement of the spectral coded target using the compensated pixel coordinate change includes: Based on the compensated pixel coordinate change, real-time focal length, pixel size of the first camera, and distance from the measuring platform to the spectral coding target, the horizontal and vertical displacements of the spectral coding target to be measured are calculated. The horizontal and vertical displacements of the spectral coding target to be measured are calculated using the following method: in, Spectral coding target Pi Horizontal displacement; Spectral coding target Pi Vertical displacement; , The amount of change in pixel coordinates after compensation; For the measurement station to the spectral coding target Pi Straight-line distance; pixel The pixel size of the first camera; f This is the real-time focal length of the first camera.

10. The long-distance, large depth-of-field displacement measurement system according to claim 1, characterized in that, The data processing device is also used for: The rate of change of pixel coordinates is verified, the abrupt change of target displacement is verified, and the correlation of multispectral coded targets is verified.

Images