Image-based suspension bridge surface anomaly measurement method and measurement system

Abstract

The invention discloses an image-based suspension bridge surface anomaly measurement method and measurement system, and relates to the technical field of optical measurement and image processing. The method comprises the following steps: projecting a first polarization grating and a second polarization grating with mutually orthogonal polarization directions to a to-be-measured area of the suspension bridge in a time-sharing manner; polarization information of reflected light is collected by adopting a polarization state imaging unit, and corresponding first and second Moire phase diagrams are generated. The data processing unit further performs double-path parallel processing on the two Moire phase diagrams: on one hand, a two-dimensional inclined field representing the macroscopic contour of an object is solved through a joint solution algorithm; and on the other hand, through differential operation, microscopic optical anisotropy abnormal points caused by material stress, damage and the like are identified and positioned by utilizing symmetrical response of the optical isotropic surface to the orthogonal grating. The method realizes synchronous decoupling of geometric and physical characteristics from a single data source through one-time measurement, has the advantages of non-contact and high efficiency, and is suitable for precise detection of large-scale structures such as a suspension bridge and the like.

Description

Technical Field

[0001] The present invention relates to the technical field of optical measurement and image processing, and in particular to an image-based surface anomaly measurement method and measurement system for a suspension bridge. Background Art

[0002] Long-term, effective health monitoring of large-scale critical infrastructure, such as suspension bridges, is crucial for ensuring their safe operation and extending their service life. This monitoring typically encompasses two equally important dimensions: macroscopic geometric posture assessment, such as measuring the tilt, deflection, and linear changes of the bridge's main cables, pylons, or deck to determine overall structural stability; and microscopic surface condition testing, aimed at promptly detecting and locating early-stage defects such as coating cracking, early corrosion, and material fatigue that could indicate structural damage.

[0003] Current technologies for acquiring macroscopic geometric information are relatively mature. Optical methods such as 3D laser scanning and photogrammetry can rapidly construct 3D point cloud models of structures, enabling accurate assessment of their overall posture and deformation. However, these techniques are primarily sensitive to the geometric contours of objects and have little ability to detect surface optical anomalies (such as birefringence) that are not accompanied by significant geometric fluctuations and are caused by material stress variations or minor physical damage. Therefore, relying solely on geometric measurements often misses the critical window for detecting early-stage structural defects.

[0004] On the other hand, a variety of technologies exist for detecting microscopic surface defects. Traditional contact methods, such as ultrasonic testing, can detect internal defects, but they are inefficient and difficult to implement over large areas. Non-contact methods such as infrared thermal imaging have applications, but their detection mechanism relies on temperature differences, limiting their applicability. More importantly, these technologies, which focus on defect detection, cannot inherently provide accurate macroscopic geometric information about the structure being tested.

[0005] As a result, the existing technology system exhibits an inherent functional separation. A comprehensive assessment of a bridge often requires deploying two or more completely different measurement systems and organizing independent measurement operations to obtain macroscopic geometric data and microscopic defect data, respectively. This separate operation mode not only leads to cumbersome and time-consuming on-site workflows, but also poses a serious technical challenge: the difficulty of data fusion. Accurately and automatically mapping local defect information obtained from one system to the global 3D geometric model obtained from another system is a nontrivial process, often accompanied by coordinate registration errors and data incompatibilities. This data-level barrier hinders the formation of a rapid, unified, and accurate holistic understanding of the structural health status.

[0006] Therefore, developing a new measurement technology that can simultaneously decouple the macroscopic geometric posture of the structure and the microscopic surface physical anomaly information from the same set of raw data in a single measurement process to overcome the bottleneck of the existing technology's single function and difficulty in data fusion has great technical value and engineering significance. Summary of the Invention

[0007] The technical problem to be solved by the present invention is that, when performing optical measurements on large structures, the existing technology usually finds it difficult to synchronously and accurately obtain the structure's macroscopic geometric posture (such as tilt) and microscopic surface physical state (such as early-stage diseases) through a single measurement, resulting in problems of single function and difficulty in data fusion.

[0008] In order to solve the above technical problems, the present invention provides a new technical solution.

[0009] A first aspect of the present invention provides an image-based method for measuring surface anomalies of a suspension bridge, the method comprising the following steps: S1. Time-sharingly projecting a first polarization grating and a second polarization grating onto a region to be measured on a suspension bridge, wherein the polarization direction of the first polarization grating varies periodically along a first spatial direction, and the polarization direction of the second polarization grating varies periodically along a second spatial direction orthogonal to the first spatial direction; S2. Using a polarization state imaging unit, respectively collecting polarization information of the first polarization grating and the second polarization grating after being reflected by the area to be measured, to obtain first polarization information and second polarization information; S3, generating a first moiré phase image and a second moiré phase image based on the first polarization information and the second polarization information respectively; S4. Process the first moiré phase image and the second moiré phase image to synchronously obtain two measurement results, the two measurement results being as follows: a. Calculate the two-dimensional tilt field characterizing the tilt state of each point on the measured area, and determine the tilt measurement results of the measured area based on the two-dimensional tilt field; b. Based on the response difference between the first moiré phase image and the second moiré phase image, identifying and locating the surface abnormal point of the test area, and determining the measurement result of the surface abnormal point.

[0010] Preferably, in step S2, the step of collecting polarization information specifically includes: The polarization state imaging unit is used to obtain intensity images I0, I1 in at least four polarization directions by one exposure. 45 ,I 90 ,I 135 ; The Stokes vectors S0, S1, S2 are calculated based on the intensity image: ; Among them, I0,I 45 ,I 90 ,I 135 are the intensity images collected by the polarization imaging unit at the polarization directions of 0°, 45°, 90°, and 135° respectively; S0, S1, and S2 are the components of the calculated Stokes vector.

[0011] Then, the linear polarization angle distribution ψ is calculated from the Stokes vector meas , to obtain the first polarization information and the second polarization information: ψ meas (x,y)=1 / 2arctan((S2(x,y)) / (S1(x,y))); Where (x, y) is the pixel coordinate in the intensity image; ψ meas (x,y) is the linear polarization angle measured at the coordinate (x,y); S1(x,y) and S2(x,y) are the component values ​​of the Stokes vector at the coordinate (x,y).

[0012] In a specific embodiment, in step S3, the step of generating a moiré phase map specifically includes: dividing the linear polarization angle distribution ψ meas The ideal reference models ψ of the first polarization grating and the second polarization grating are ref Perform a differential operation to generate the wrapped moiré phase map ΔΦ: ΔΦ(x,y)=ψ meas (x,y)-ψ ref (x,y); Where ΔΦ(x,y) is the moiré phase value generated at the coordinate (x,y); ψ ref (x,y) is the theoretical reference linear polarization angle of the projected polarization grating at coordinate (x,y).

[0013] Furthermore, in step S4, the step of determining the tilt measurement result of the area to be measured includes: according to the pre-calibrated system transfer function matrix C, the first moiré phase image ΔΦ H and the second moiré phase pattern ΔΦ V Perform joint solution to obtain the tilt component θ x and θ y The two-dimensional tilt field formed: ; Among them, θ x (x,y) and θ y (x, y) are the tilt components of the surface of the measured area at the coordinate (x, y) along the first spatial direction and the second spatial direction respectively; C(x, y) is the 2x2 system transfer function matrix pre-calibrated at the coordinate (x, y); ΔΦH (x, y) is the phase value of the first moiré phase pattern generated by the first polarization grating at the coordinate (x, y); ΔΦ V (x, y) is the phase value of the second moiré phase pattern generated by the second polarization grating at the coordinate (x, y).

[0014] The two-dimensional tilt field is converted into a visual cloud map or a vector arrow map as the tilt measurement result.

[0015] In one embodiment, the joint solution is specifically: The integrability condition that the two-dimensional tilt field must satisfy is used as an optimization constraint. The first and second moiré phase images are solved by an optimization algorithm to obtain the two-dimensional tilt field. The calculation error caused by optical distortion or component misalignment is corrected based on the two-dimensional tilt field.

[0016] Preferably, in step S4, the step of determining the measurement result of the surface abnormal point includes: H and the second moiré phase pattern ΔΦ V Perform differential operation to generate differential anomaly map M diff Optically isotropic points in the differential anomaly map have a symmetrical response to the orthogonal polarization grating, while optically anisotropic points exhibit an asymmetric response. The differential operation amplifies the asymmetric response, thereby forming a recognizable signal in the differential anomaly map. Subsequently, points in the differential anomaly map with values ​​higher than a preset threshold are identified as surface anomalies, and their position information is output as the measurement result of the surface anomaly.

[0017] In a specific embodiment, the differential operation specifically includes: normalizing the first moiré phase image and the second moiré phase image, and calculating the difference between the normalized first moiré phase image and the second moiré phase image to generate the differential anomaly image M diff : M diff (x,y)=|(ΔΦ H (x,y)) / σ H -(ΔΦ V (x,y)) / σ V |; Among them, M diff (x,y) is the value of the differential anomaly map at the coordinate (x,y); σ H and σ v are the local standard deviations of the first and second moiré phase images, respectively, used for normalization.

[0018] Preferably, in step S1, a spatial light modulator is used to generate and switch the first polarization grating and the second polarization grating.

[0019] In a specific embodiment, the first spatial direction is a horizontal direction, and the second spatial direction is a vertical direction.

[0020] A second aspect of the present invention provides an image-based suspension bridge surface anomaly measurement system, the system comprising: a projection unit, configured to project a first polarization grating and a second polarization grating onto the area to be measured of the suspension bridge in a time-sharing manner, wherein the polarization direction of the first polarization grating varies periodically along a first spatial direction, and the polarization direction of the second polarization grating varies periodically along a second spatial direction orthogonal to the first spatial direction; a polarization state imaging unit, configured to respectively collect polarization information of the first polarization grating and the second polarization grating after being reflected by the area to be measured, to obtain first polarization information and second polarization information; A data processing unit is connected to the projection unit and the polarization state imaging unit, and is used to generate a first moiré phase map and a second moiré phase map based on the first polarization information and the second polarization information, respectively, and process the first moiré phase map and the second moiré phase map to synchronously obtain the tilt measurement results of the measured area and the measurement results of the surface abnormal points.

[0021] The present invention provides an image-based method and system for measuring surface anomalies of a suspension bridge. This method has the following beneficial effects: 1. The present invention uses time-sharing projection of a first polarization grating and a second polarization grating to obtain first and second moiré phase images. On the one hand, the two-dimensional tilt field of the measured area is determined by joint solution, and on the other hand, surface anomalies are identified through differential calculation. This solution integrates macroscopic geometric posture measurement with microscopic surface state detection into the same data acquisition and processing flow. A single measurement can simultaneously output detection results of two different physical dimensions, without the need for additional hardware equipment or independent measurement operations.

[0022] 2. The technical solution of the present invention generates two moiré phase maps containing data redundancy by collecting independent information generated by two polarization gratings in orthogonal directions. In the process of jointly solving the tilt component, this data redundancy can be used to compensate for and suppress the inherent errors in the system, thereby realizing self-calibration of the measurement. This design reduces the system's sensitivity to environmental changes and hardware instability, ensuring the accuracy of the measurement results.

[0023] 2. The measurement system of the present invention uses a single polarization state imaging unit and can complete the acquisition of two orthogonal component information by quickly switching the projected grating. This process does not require complex mechanical movement or system reconstruction. In particular, when a spatial light modulator is used to generate and switch the grating, the data acquisition speed is fast and the operation process is simple, which significantly shortens the on-site operation time. It is more suitable for rapid, in-situ detection of large structures such as suspension bridges. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 Schematic diagram of the structure of the image-based suspension bridge surface anomaly measurement system of the present invention; Figure 2 This is a schematic structural diagram of a projection unit according to an embodiment of the present invention; Figure 3 Schematic diagram of the structure of a polarization imaging unit according to an embodiment of the present invention; Figure 4 is a flow chart of the image-based suspension bridge surface anomaly measurement method of the present invention; Figure 5 A schematic diagram of a standard test sample according to an embodiment of the present invention; Figure 6 A two-dimensional tilt field cloud image according to an embodiment of the present invention; Figure 7 This is a differential anomaly map according to an embodiment of the present invention.

[0025] Reference numerals in the accompanying drawings: 10, projection unit; 11, light source; 12, collimation and beam expansion system; 13, polarizer; 14, spatial light modulator; 15, projection lens; 20, polarization state imaging unit; 21, imaging lens; 22, micro-polarizer array; 23, image sensor; 30, data processing unit. DETAILED DESCRIPTION

[0026] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the present specification. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0027] Refer to the attached Figure 1 This embodiment discloses a measurement system for implementing the image-based suspension bridge surface anomaly measurement method. The measurement system may include: a projection unit 10, a polarization state imaging unit 20, and a data processing unit 30.

[0028] The projection unit 10 is used for projecting the first polarization grating and the second polarization grating onto the area to be measured S of the suspension bridge in a time-sharing manner.

[0029] The polarization state imaging unit 20 is used to collect polarization information of the reflected light after being reflected by the area S to be measured.

[0030] The data processing unit 30 establishes communication connections with the projection unit 10 and the polarization imaging unit 20. The data processing unit 30 sends instructions to the projection unit 10 to control the generation and projection timing of the grating, and simultaneously sends trigger signals to the polarization imaging unit 20 to synchronize the acquisition of raw image data. The polarization imaging unit 20 transmits the acquired raw image data to the data processing unit 30, which performs subsequent calculations and analysis tasks and outputs the final measurement results.

[0031] Refer to the attached Figure 2 In a specific embodiment, the projection unit 10 includes, along the optical path, a light source 11 , a collimating beam expansion system 12 , a polarizer 13 , a spatial light modulator 14 and a projection lens 15 .

[0032] Light source 11 provides a diverging beam, such as a light-emitting diode (LED) or laser. The output diverging beam is either unpolarized or randomly polarized. A collimating and expanding system 12, located after light source 11 and consisting of a set of lenses, shapes the diverging beam emitted from light source 11 into a parallel beam with a predetermined aperture, ensuring complete coverage of the test area S of the suspension bridge.

[0033] Polarizer 13 is positioned between collimating and expanding system 12 and spatial light modulator 14. Its transmission axis is oriented along a predetermined direction, such as horizontal. After passing through polarizer 13, the parallel light beam is converted into linearly polarized light, providing a reference polarization state for the subsequent spatial light modulator.

[0034] The spatial light modulator 14, for example, a liquid crystal spatial light modulator (LC-SLM), is a component used to generate a polarization grating. The spatial light modulator 14 is connected to the data processing unit 30 and receives the digital grayscale image generated by the data processing unit 30. Each pixel of the spatial light modulator 14 independently rotates the polarization direction of the linearly polarized light passing through it based on the received grayscale value, thereby forming a two-dimensional light field in the cross section of the outgoing light beam, with the polarization direction spatially distributed according to a predetermined pattern, i.e., a polarization grating.

[0035] Specifically, to generate the first polarization grating, the data processing unit 30 generates a digital grayscale image whose grayscale values ​​periodically vary along a first spatial direction (e.g., the horizontal x-direction) and loads it into the spatial light modulator 14. Based on this, the spatial light modulator 14 generates a first polarization grating whose polarization direction periodically and continuously varies in the horizontal direction. Similarly, to generate the second polarization grating, the data processing unit 30 generates another digital grayscale image whose grayscale values ​​periodically vary along a second spatial direction orthogonal to the first spatial direction (e.g., the vertical y-direction) and loads it into the spatial light modulator 14, thereby generating a second polarization grating whose polarization direction periodically varies in the vertical direction.

[0036] Data processing unit 30 controls the switching of the digital grayscale image loaded into spatial light modulator 14, thereby achieving time-sharing projection of the first and second polarization gratings. Projection lens 15, located at the end of the optical path, is used to image the light field output by spatial light modulator 14, which carries the polarization grating information, onto the surface of the test area S.

[0037] Refer to the attached Figure 3 , Figure 3 1 is a schematic structural diagram of a polarization imaging unit according to an embodiment of the present invention.

[0038] In a preferred embodiment, the polarization state imaging unit 20 is a “snapshot” split-focus plane polarization camera, which mainly includes an imaging lens 21 , a micro-polarizer array 22 , and an image sensor 23 .

[0039] The imaging lens 21 is used to focus and form an image of the light field reflected by the to-be-measured area S. The image sensor 23 may be a charge coupled device (CCD) or a complementary metal oxide semiconductor (CMOS) sensor for photoelectric conversion.

[0040] The micropolarizer array 22 is the core component of this "snapshot" focal plane polarization camera. It is integrated and precisely aligned on the surface of the photosensitive pixel array of the image sensor 23. The micropolarizer array 22 is composed of a large number of repeated superpixel units, each of which covers, for example, a 2x2 grouping of sensor pixels. Within each superpixel unit, four micropolarizers have polarization directions of 0°, 45°, 90°, and 135°, respectively, corresponding one-to-one to the four sensor pixels below.

[0041] When the reflected light from the test area S is imaged onto the micro-polarizer array 22 via the imaging lens 21, the light passing through a single superpixel unit is decomposed in real time into four components with different polarization directions before entering the sensor pixel below. Therefore, during a single exposure, different sensor pixels on the image sensor 23 simultaneously record the light intensity distribution after being polarized at four different angles. This "snapshot" focal plane polarization camera outputs a raw mosaic image containing all polarization information, which is then transmitted to the data processing unit 30. The data processing unit 30 uses a specific demosaicing algorithm to reconstruct four complete, aligned intensity images corresponding to polarization directions of 0°, 45°, 90°, and 135° from the raw mosaic image at one time. The raw mosaic image is the raw image data.

[0042] In another embodiment, the polarization state imaging unit 20 can also consist of a standard industrial camera and a rotatable polarizer placed in front of its imaging lens. The rotatable polarizer is driven by a precision motor connected to the data processing unit 30. During measurement, the data processing unit 30 controls the precision motor to rotate the polarizer to 0°, 45°, 90°, and 135°, triggering the standard industrial camera to perform an exposure in each polarization direction, thereby acquiring four intensity images in a time-sharing manner.

[0043] Refer to the attached Figure 1 The data processing unit 30 is the control and computing component of the entire measurement system. This unit can be an industrial personal computer, a portable computer, or an embedded system with an integrated processor and memory. It includes a central processing unit (CPU), a graphics processing unit (GPU), memory (RAM), data storage devices (such as solid-state drives), and communication interfaces for connecting to external devices.

[0044] The data processing unit 30 is connected to the projection unit 10 and the polarization imaging unit 20 via communication interfaces. For example, it transmits a digital grayscale image to the spatial light modulator 14 in the projection unit 10 via a digital video interface (such as HDMI) and sends a trigger signal to the polarization imaging unit 20 via a general-purpose input / output (GPIO) or a dedicated communication bus. The polarization imaging unit 20 transmits the acquired raw image data to the data processing unit 30 via a high-speed data interface (such as USB 3.0 or GigE Vision).

[0045] The data processing unit 30 runs proprietary measurement and analysis software, whose functional modules are responsible for executing the measurement method of the present invention. The main functions include: System control and synchronization: Generate and send digital grayscale images for forming the first polarization grating and the second polarization grating to the projection unit 10 according to a predetermined timing; during each grating projection, synchronously send acquisition instructions to the polarization state imaging unit 20 to ensure accurate timing matching of data acquisition.

[0046] Data reception and preprocessing: Receives raw image data transmitted from the polarization imaging unit 20. This data is decoded to obtain intensity images for four polarization directions (0°, 45°, 90°, and 135°). The Stokes vector components corresponding to each pixel coordinate in the intensity image are calculated.

[0047] Core algorithm execution: Based on the calculated Stokes vector, the linear polarization angle distribution is further calculated. By differentiating the linear polarization angle distribution with a pre-stored ideal reference model, the first and second moiré phase maps are generated. Subsequently, a joint solution algorithm and a differential operation are performed to obtain a two-dimensional tilt field and a differential anomaly map representing the measured area, respectively.

[0048] Result Generation and Output: Convert the calculated 2D tilt field data into a visual cloud map or vector arrow plot. Threshold processing is performed on the differential anomaly map to extract the location information of surface anomalies, which is used as the surface anomaly measurement result. Finally, the tilt measurement results and surface anomaly measurement results are displayed on the user interface or saved as a data file in a specified format.

[0049] The principle of the tilt measurement result of the present invention is based on the modulation effect of the local tilt of the surface to be measured on the polarization state of the reflected light.

[0050] When a beam of linearly polarized light is incident on a tiny surface element in the measurement area S, if the surface element is an ideal Lambertian body, the polarization state of the reflected light changes. This change is related to the geometric relationship between the surface element's normal vector n and the direction of the incident light and the observation direction. Therefore, by accurately measuring the change in the polarization state of the reflected light, the normal direction of the tiny surface element, that is, its tilt state, can be inferred.

[0051] In this embodiment, the projection unit 10 first projects a first polarization grating, whose polarization direction varies periodically along a first spatial direction (e.g., the x-axis). In an ideal case where the influence of surface tilt is not considered, the reference linear polarization angle distribution ψ on the imaging plane is ref,H (x, y) is known. When the grating is projected onto the area to be measured S with a local tilt, the linear polarization angle distribution measured by the polarization state imaging unit 20 is ψ meas,H (x, y). The first moiré phase ΔΦ H (x,y) is defined as the linear polarization angular distribution ψ meas,H (x, y) and the reference linear polarization angle distribution ψref,H The difference between (x, y), the first moiré phase contains the information of surface tilt.

[0052] Similarly, the projection unit 10 projects a second polarization grating. When the polarization direction of the second polarization grating changes periodically along a second spatial direction (eg, the y-axis) orthogonal to the first spatial direction, a second moiré phase ΔΦ can be obtained. V (x,y).

[0053] Moiré phase ΔΦ H (x,y) and ΔΦ V (x, y) and the two orthogonal tilt components θ at the corresponding point of the coordinate (x, y) of the area to be measured S x and θ y There is a linear relationship between them. x =∂z / ∂x and θ y =∂z / ∂y, where z(x,y) is the height distribution of the surface to be measured. This relationship can be described by a system of two linear equations.

[0054] Writing this system of equations in matrix form, we can get: ; Among them, M(x,y) is a 2x2 coefficient matrix. The elements in M(x,y) are determined by the geometric configuration of the measurement system (such as illumination angle, observation angle, etc.) and are different at each pixel coordinate (x,y).

[0055] By solving the above matrix equation, the tilt component θ corresponding to each pixel coordinate can be uniquely calculated: x and θ y This is achieved by calculating the inverse matrix of the matrix M(x,y), which is the pre-calibrated system transfer function matrix C(x,y).

[0056] Therefore, the calculation formula of the tilt component can be clearly expressed as: ; Among them, θ x (x,y) and θ y (x, y) are the tilt components of the surface of the measured area at the coordinate (x, y) along the first spatial direction and the second spatial direction; ΔΦ H (x, y) is the phase value of the first moiré phase pattern generated by the first polarization grating at the coordinate (x, y); ΔΦ V (x, y) is the phase value of the second moiré phase pattern generated by the second polarization grating at the coordinate (x, y); C(x, y) is the 2x2 system transfer function matrix pre-calibrated at the coordinate (x, y).

[0057] By performing this calculation on all pixel coordinates in the first and second moiré phase images, a complete two-dimensional tilt field can be obtained, thereby characterizing the macroscopic tilt state of the entire measured area.

[0058] The principle of synchronously detecting surface abnormalities in the present invention is that there is an essential difference in the symmetry of the responses of the intact area and the abnormal area within the test area S to orthogonal polarized light.

[0059] For a physically and chemically homogeneous, optically isotropic intact area, such as an intact, uniform paint coating, its reflection characteristics for incident light are independent of its polarization direction. Therefore, when two polarization gratings (a first polarization grating and a second polarization grating) with mutually orthogonal polarization directions are projected onto the intact area in a time-sharing manner, the orthogonal polarization light responses generated in the two measurement channels are symmetrical. This means that the first moiré phase pattern ΔΦ generated by the first polarization grating and the second polarization grating, respectively, is symmetrical. H and the second moiré phase pattern ΔΦ V The local variation characteristics of the orthogonal polarization light response in this intact area should be consistent, except for the inherent structural differences caused by different grating directions.

[0060] However, when certain early-stage defects or defects appear on the surface of a suspension bridge, such as microscopic rust on the paint coating, surface scratches, water stains, or material deterioration, these abnormal areas can change their microstructure, causing them to exhibit optical anisotropy. An optically anisotropic surface is characterized by its interaction with light (such as reflection and scattering) depending on the polarization direction of the incident light.

[0061] When this type of optical anisotropy anomaly exists on the test area S, the symmetry of the orthogonal polarization response is destroyed. The reflection response of the anomaly to the first polarization grating with a polarization direction changing along the first spatial direction is different from the reflection response of the anomaly to the second polarization grating with a polarization direction changing along the second spatial direction. This difference will be directly reflected in the moiré phase pattern ΔΦ H and ΔΦ V In this way, the abnormal point generates an asymmetric local phase change at the pixel coordinates corresponding to the first moiré phase image and the second moiré phase image.

[0062] To identify and locate this asymmetry, the present invention performs a differential operation on the two moiré phase images. This operation aims to suppress the symmetrical common-mode signal components in the two moiré phase images due to macroscopic surface tilt, while amplifying the asymmetric differential-mode signal components due to local optical anisotropy. In a specific embodiment, this differential operation is implemented using the following formula: M diff (x,y)=|(ΔΦH (x,y)) / σ H -(ΔΦ V (x,y)) / σ V |; Among them, M diff (x,y) is the value of the differential anomaly map at the coordinate (x,y); ΔΦ H (x,y) and ΔΦ V (x,y) are the phase values ​​of the first and second moiré phase patterns at coordinates (x,y); σ H and σ V are the local standard deviations of the first and second moiré phase images in a local neighborhood of coordinate (x, y), respectively.

[0063] In this operation, the local standard deviation σ H and σ V It plays the role of a normalization factor. By dividing the phase value by its local standard deviation, the difference in the phase change amplitude caused by the macroscopic geometric shape of the surface can be eliminated, so that the differential operation can more sensitively reflect the response asymmetry caused by the difference in the optical properties of the material itself. The differential anomaly map M obtained after calculation diff In the graph, the areas with values ​​close to zero correspond to optically isotropic intact surfaces, while the areas with values ​​significantly higher than the background indicate the locations of optically anisotropic anomalies.

[0064] The measurement method of the present invention begins with step S1 , ie, generating and projecting a polarization grating. This step is performed by the projection unit 10 under the control of the data processing unit 30 .

[0065] First, to generate the first polarization grating, the data processing unit 30 internally calculates and generates a first digital grayscale image. In this first digital grayscale image, the grayscale value of a pixel coordinate is a periodic function of its position in a first spatial direction (e.g., the x-axis direction in the image coordinate system). For example, for an 8-bit first digital grayscale image, the grayscale value G at the coordinate (x, y) is H (x,y) can be represented by G H (x,y)=mod(ax,256), where a is the coefficient that determines the spatial frequency of the grating and mod is the modulo operation.

[0066] The data processing unit 30 transmits this first digital grayscale image to the spatial light modulator 14 in the projection unit 10 via a data interface. The spatial light modulator 14 receives this first digital grayscale image. Based on the grayscale values ​​received by each pixel, the liquid crystal cells within the spatial light modulator 14 apply a precise polarization rotation corresponding to the grayscale value to the incident uniform linearly polarized light, which has been pre-processed by the polarizer 13. As a result, the polarization direction of the cross-section of the light field emitted from the spatial light modulator 14 exhibits a predetermined periodic variation in the first spatial direction, thereby forming a first polarization grating. This first polarization grating is then projected onto the area to be measured S via the projection lens 15.

[0067] After completing the image acquisition corresponding to the first polarization grating, the data processing unit 30 generates a second digital grayscale image. In the second digital grayscale image, the grayscale value of the pixel coordinate is a periodic function of its position in the second spatial direction (e.g., the y-axis direction in the image coordinate system, which is orthogonal to the first spatial direction). The grayscale value G of the second digital grayscale image at the coordinate (x, y) is V (x,y) can be represented by G V (x,y)=mod(by,256), where b is the coefficient that determines the spatial frequency of the grating.

[0068] The data processing unit 30 loads this second digital grayscale image into the spatial light modulator 14, replacing the first digital grayscale image. In the same manner, the spatial light modulator 14 generates a second polarization grating whose polarization direction periodically varies along a second spatial direction and projects it onto the same measurement area S. In this way, the system achieves time-sharing projection of the first and second polarization gratings.

[0069] In another embodiment, the projection unit 10 can also be implemented using a digital light processing (DLP) projector. In this case, the DLP chip acts as a spatial light modulator. Since standard DLP projectors do not inherently modulate polarization states, a motor-driven, synchronously rotating filter wheel containing multiple polarization filters in different orientations can be added behind the projection lens 15. When the DLP projects an intensity image corresponding to a first grating, the data processing unit 30 synchronously controls the filter wheel to place a polarizer with a specific orientation (for example, a liquid crystal micro-grating with a horizontally varying polarization direction) in the light path to form a first polarization grating. When switching to the second grating, the DLP loads the new intensity image, and the filter wheel simultaneously rotates to the corresponding second polarizer. This approach can also achieve time-sharing projection of polarization gratings.

[0070] Refer to the attached Figure 4 , Figure 4 This is a flow chart of the image-based suspension bridge surface anomaly measurement method of the present invention. The above system part can realize the technical content of the following method part, which is as follows: When the first polarization grating is projected in step S1 , the data processing unit 30 synchronously sends a trigger instruction to the polarization state imaging unit 20 to execute the acquisition of polarization information.

[0071] The polarization state imaging unit 20 images the first polarization grating reflected by the area to be measured S. In a preferred embodiment, when a "snapshot" focal plane polarization camera is used, the camera captures an original mosaic image in one exposure. This original mosaic image is immediately transmitted to the data processing unit 30. The data processing unit 30 performs a demosaicing algorithm on the original mosaic image. This algorithm interpolates and reconstructs the original mosaic image based on the known directions of the four micro-polarizers in each 2x2 super-pixel unit, thereby generating four intensity images with one-to-one correspondence and the same resolution: I0(x, y), I 45 (x,y),I 90 (x,y) and I 135 (x,y).

[0072] The data processing unit 30 then uses the four intensity images to calculate the three components S0, S1, S2 of the Stokes vector for each pixel coordinate (x, y) in the four intensity images: ; Subsequently, the data processing unit 30 further calculates the linear polarization angle distribution ψ measured at each pixel coordinate based on the calculated Stokes vector component. meas (x, y), this distribution is the first polarization information: ψ meas,H (x, y)=1 / 2arctan((S2(x, y)) / (S1(x, y))), where the subscript H indicates that the first polarization information is generated by a first polarization grating (eg, a horizontal grating).

[0073] After completing the acquisition of the first polarization information, when step S1 switches and projects the second polarization grating, the above acquisition and calculation process is completely repeated. The polarization state imaging unit 20 acquires the second polarization grating reflected by the test area S, and the data processing unit 30 processes the received raw image data in exactly the same way, and finally generates the linear polarization angle distribution ψ corresponding to the second polarization grating. meas,V (x, y), this distribution is the second polarization information.

[0074] In step S2, the linear polarization angle distribution ψ is obtained. meas Then, the method proceeds to step S3, where the data processing unit 30 generates a moiré phase map. This step aims to extract the phase deviation caused by the geometric shape and surface characteristics of the measured area S from the measurement data.

[0075] The data processing unit 30 pre-stores two ideal reference models: the first reference model ψ ref,H (x,y) and the second reference model ψ ref,V (x,y). First reference model ψ ref,H (x, y) is the theoretical linear polarization angle distribution that should be formed at the polarization state imaging unit 20 when the first polarization grating is projected on an ideal plane without tilt and defects. Similarly, the second reference model ψ ref,V (x,y) is the theoretical linear polarization angle distribution corresponding to the second polarization grating. These two reference models can be generated by mathematical calculation or obtained by a one-time calibration measurement on a standard flat sample.

[0076] To generate the first moiré phase image, the data processing unit 30 processes the first polarization information ψ obtained in step S2. meas,H (x,y) and the pre-stored first reference model ψ ref,H (x, y) performs pixel-by-pixel difference operation, and the process follows the following relationship: ΔΦ H (x,y)=ψ meas,H (x,y)-ψ ref,H (x,y) The result of this operation ΔΦ H (x,y) is the first moiré phase image.

[0077] Similarly, the data processing unit 30 processes the second polarization information ψ meas,V (x, y) and the pre-stored second reference model ψ ref,V The same pixel-by-pixel difference operation is performed on (x, y) to generate the second moiré phase map ΔΦ V (x,y):ΔΦ V (x,y)=ψ meas,V (x,y)-ψ ref,V (x,y).

[0078] Due to ψ meas The calculation involves the inverse tangent function, whose value is limited to an interval of π, so the moiré phase map ΔΦ generated by the above difference operation is H and ΔΦ V The value range of is limited to (-π, π], and this phase diagram is called a wrapped Moiré phase diagram. Before entering the next step, the data processing unit 30 performs a phase unwrapping algorithm on the two wrapped Moiré phase diagrams to eliminate the 2π phase jump, thereby restoring a continuous phase distribution that can truly reflect the phase change, providing a data basis for subsequent tilt field solution and outlier identification.

[0079] After the continuous moiré phase images are generated in step S3, the data processing unit 30 performs step S4 to process the first moiré phase image ΔΦ. H(x,y) and the second moiré phase pattern ΔΦ V (x, y) is processed to synchronously obtain the tilt measurement results of the area to be measured and the measurement results of the surface abnormal points.

[0080] To obtain the tilt measurement results of the measured area, the data processing unit 30 jointly solves the two moiré phase images. This solution is based on a system transfer function matrix C(x,y) that has been previously obtained through a calibration process and stored in the data processing unit 30. For each pixel coordinate (x,y) in the moiré phase image, the data processing unit 30 performs the following matrix operation to solve for the two-dimensional tilt component θ of each pixel coordinate (x,y): x and θ y : ; By performing this calculation for all pixel coordinates within the field of view, data processing unit 30 obtains a complete two-dimensional tilt field consisting of the tilt components of all pixel coordinates. Finally, data processing unit 30 visualizes this two-dimensional tilt field data, for example, rendering it as a cloud map whose color varies with the tilt amplitude, or as a vector arrow map whose arrow direction and length indicate the tilt direction and magnitude. This cloud map and vector arrow map are then output or displayed as the tilt measurement results.

[0081] At the same time, the data processing unit 30 uses the response difference between the first Moiré phase image and the second Moiré phase image to determine the measurement result of the surface anomaly point. The data processing unit 30 performs a differential operation on the two unwrapped Moiré phase images to generate a differential anomaly map M diff (x, y). In a specific embodiment, the operation specifically includes performing local normalization and difference calculation on the first moiré phase image and the second moiré phase image: M diff (x,y)=|(ΔΦ H (x,y)) / σ H -(ΔΦ V (x,y)) / σ V |, where σ H and σ V are the standard deviations of the first and second moiré phase patterns within a local neighborhood centered at coordinate (x,y). This normalization step aims to eliminate variations in the response amplitude caused by the macroscopic geometry of the surface, thereby increasing sensitivity to asymmetric responses caused by optical anisotropy of the surface material.

[0082] Generate differential anomaly map M diffAfterwards, the data processing unit 30 compares the value with a preset numerical threshold. In the differential anomaly map, pixel coordinates with values ​​above the threshold are identified as surface outliers. Finally, the data processing unit 30 extracts the position information of all identified surface outliers and outputs this position information as the surface outlier measurement results, or overlays it on the tilt measurement result map.

[0083] In a preferred embodiment, in order to ensure the accuracy of the measurement results, the method of the present invention further includes a system calibration step performed before formal measurement, and a self-calibration step performed during data processing.

[0084] The purpose of the system calibration process is to accurately determine the system transfer function matrix C(x,y) used in step S4. This process uses a high-precision flat calibration plate mounted on a two-dimensional turntable that can be accurately rotated along two orthogonal axes (x and y). First, the calibration plate is placed in a zero position perpendicular to the optical axis (i.e., θ x =0,θ y =0). Then, the calibration plate is precisely rotated around the y-axis by a known small angle Δθ. x , and a small known angle Δθ around the x-axis y In each posture, the complete steps S1 to S3 are executed to obtain the corresponding Moiré phase map. x ,Δθ y ) and the resulting moiré phase pattern variation, the corresponding transfer function matrix C(x,y) can be inversely solved for each coordinate (x,y) in the field of view. This matrix is ​​digitally stored in the data processing unit 30 for subsequent measurement.

[0085] More specifically, the calibration process is as follows: Place the calibration plate in zero position (θ x =0,θ y =0), collect and calculate a set of reference Moiré phase images ΔΦ H,0 and ΔΦ V,0 .

[0086] The calibration plate is precisely rotated around the y-axis by a known small angle Δθ x , at this time the posture is (θ x =Δθ x ,θ y =0). Collect and calculate the moiré phase map ΔΦ again H,1 and ΔΦ V,1 .

[0087] Return the calibration plate to zero position, and then precisely rotate it around the x-axis by a known small angle Δθy , at this time the posture is (θ x =0,θ y =Δθ y ). Collect and calculate the Moiré phase map ΔΦ again H,2 and ΔΦ V,2 .

[0088] According to the linear relationship in step S4, for each pixel coordinate (x, y), we have: ; ; From this, the inverse matrix C of the transfer function matrix can be solved point by point -1 By performing this operation on all pixels in the field of view, we can obtain the complete system transfer function matrix.

[0089] The above two formulas are used to solve the transfer function matrix C (or its inverse matrix C -1 ) specific mathematical steps. Our goal is to find C -1 All four unknown elements of this 2x2 matrix.

[0090] First, C -1 Written in the form of specific elements: ; For the first formula: Physical operation: We rotate the calibration plate around the y-axis by a known angle Δθ x , while the rotation around the x-axis remains zero.

[0091] Input: The tilt change corresponding to this operation is a vector .

[0092] Output: We measured that the Moiré phase image also produced a change compared to the zero-position posture. This change is the vector .

[0093] Mathematical relationship: Substituting these into (phase change) = C -1 (angle change), we get: ; Solution: From this matrix multiplication we can see that ΔΦ V,1 -ΔΦ V,0 =c' 11 Δθ x And ΔΦ H,1 -ΔΦ H,0 =c' 21 Δθ xSince the phase change on the left side of the equal sign is measured, Δθ x is known, so we can directly calculate c' 11 and c' 21 These two values. This is equivalent to us determining C -1 The first column of the matrix.

[0094] For the second formula: Physical operation: Similarly, we rotate the calibration plate around the x-axis by a known angle Δθ y , while the rotation around the y-axis remains at 0.

[0095] Input: The tilt change is .

[0096] Output: The measured phase change is .

[0097] Mathematical relationship:

[0098] Solution: From this, we can calculate c' 12 and c' 22 These two values. This is equivalent to us determining C -1 The second column of the matrix.

[0099] Through these two steps, we have successfully determined C -1 All four elements of the matrix. Finally, -1 By taking the inverse once, we get the system transfer function matrix C that we ultimately need.

[0100] The self-calibration process is performed in the joint solution process of step S4, which uses the data redundancy provided by the two measurements (the first polarization grating and the second polarization grating) to compensate for the systematic error. The physical basis of this process is that the two-dimensional tilt field (θ x ,θ y ) must satisfy the integrability condition, that is, its mixed partial derivatives must be equal: ∂θ x / ∂y=∂θ y / ∂x. In actual measurements, the calculated tilt field caused by system errors (such as optical distortion or slight component misalignment) may not fully meet this condition. Therefore, when performing the solution, the data processing unit 30 can use the satisfaction of the integrability condition as a constraint and adjust the solution results through an optimization algorithm (such as the least squares method) or fine-tune the transfer function matrix C(x,y) to minimize the integrability error of the final output two-dimensional tilt field. This process uses the measurement data itself to correct the system model in real time, thereby compensating for slight drifts in the system state and further improving the accuracy and reliability of the measurement results.

[0101] Please refer to the attached Figure 5-7 , which shows the scene and measurement results of using the system of the present invention to detect a standard test sample.

[0102] In this embodiment, the measurement target is a specially prepared standard test sample T, as shown in the attached Figure 6 The sample consists of a flat metal substrate with a smooth cylindrical protrusion machined into its center to simulate macroscopic geometric shapes. Furthermore, to simulate microscopic surface anomalies, an extremely thin birefringent film with known optical anisotropy is pre-attached to a flat area of ​​the sample substrate.

[0103] The measurement system is placed about 1 meter in front of the standard test sample T. The operator selects the area to be measured that includes the entire sample through the control software interface of the data processing unit 30.

[0104] After the measurement starts, the system automatically executes the following process: The projection unit 10 first projects a first polarization grating (the polarization direction of which changes periodically along the horizontal direction).

[0105] The polarization state imaging unit 20 synchronously collects a piece of original image data.

[0106] The projection unit 10 switches and projects a second polarization grating (the polarization direction of which changes periodically along the vertical direction).

[0107] The polarization state imaging unit 20 collects the second raw image data. The entire data collection process takes less than 0.1 seconds. The data processing unit 30 then automatically processes the two collected raw image data and simultaneously outputs two measurement results.

[0108] a. Obtaining and analyzing the tilt measurement results The two-dimensional tilt field cloud generated by the data processing unit 30 is displayed on the screen, as shown in the attached Figure 7 This contour map clearly reflects the macroscopic geometry of the standard test specimen T. It can be observed that the specimen's flat base region exhibits a uniform slope close to zero (shown in blue), while the cylindrical protrusion exhibits a smooth, radial gradient around its center, accurately reproducing the known geometric profile of the specimen.

[0109] b. Synchronous detection and verification of surface anomalies At the same time, the differential anomaly map generated by the data processing unit 30 is also output, as shown in the attached figure. Figure 7 As shown. Figure 7 In the vast majority of the area (corresponding to the uniform metal surface), the value is close to zero and appears as a dark background. Figure 7A bright spot appears at one location in the image, representing a bright abnormality. The location of this bright abnormality perfectly matches the location of an ultrathin birefringent film previously applied to the sample surface. This result strongly demonstrates the high sensitivity of the method presented herein for detecting optical anisotropy caused by physical properties such as birefringence, enabling reliable identification of microscopic surface anomalies.

[0110] This embodiment fully demonstrates that the measurement system and method provided by the present invention can simultaneously obtain macroscopic three-dimensional geometric information and microscopic physical property anomaly information of the surface of the object to be measured through a single rapid and non-contact measurement, thereby realizing the efficient integration of multimodal detection functions and having extremely high engineering and scientific research application value.

Claims

1. A method for measuring surface anomalies of a suspension bridge based on an image, characterized in that: The following steps are involved: S1. Time-sharingly projecting a first polarization grating and a second polarization grating onto a region to be measured on a suspension bridge, wherein the polarization direction of the first polarization grating varies periodically along a first spatial direction, and the polarization direction of the second polarization grating varies periodically along a second spatial direction orthogonal to the first spatial direction; S2. Using a polarization state imaging unit, respectively collecting polarization information of the first polarization grating and the second polarization grating after being reflected by the area to be measured, to obtain first polarization information and second polarization information; S3. Generate a first moiré phase image and a second moiré phase image based on the first polarization information and the second polarization information respectively; S4. Process the first moiré phase image and the second moiré phase image to synchronously obtain two measurement results, the two measurement results being as follows: a. Calculating a two-dimensional tilt field characterizing the tilt state of each point on the measured area, and determining the tilt measurement results of the measured area based on the two-dimensional tilt field; b. Based on the response difference between the first moiré phase image and the second moiré phase image, identifying and locating the surface abnormal point of the test area, and determining the measurement result of the surface abnormal point.

2. The method for measuring surface anomalies of a suspension bridge based on an image according to claim 1, characterized in that: In step S2, collecting polarization information using the polarization state imaging unit specifically includes: Obtain intensity images in at least four polarization directions by a single exposure using the polarization state imaging unit; Calculating polarization information including a Stokes vector based on the intensity images of the at least four polarization directions; The linear polarization angle distribution is calculated from the Stokes vector to obtain the first polarization information and the second polarization information.

3. The method for measuring surface anomalies of a suspension bridge based on an image according to claim 2, characterized in that: In step S3, generating the first moiré phase image and the second moiré phase image specifically includes: The linear polarization angle distribution calculated from the first polarization information and the second polarization information is subjected to differential operation with the ideal reference models of the first polarization grating and the second polarization grating, respectively, to generate the first moiré phase image and the second moiré phase image.

4. The method for measuring surface anomalies of a suspension bridge based on an image according to claim 1, characterized in that: In step S4, determining the tilt measurement result of the area to be measured specifically includes: performing a joint solution on the first moiré phase map and the second moiré phase map according to a pre-calibrated system transfer function matrix to obtain the two-dimensional tilt field; The two-dimensional tilt field is converted into a visual cloud map or a vector arrow map as the tilt measurement result.

5. The method for measuring surface anomalies of a suspension bridge based on an image according to claim 1, characterized in that: In step S4, determining the measurement result of the surface abnormal point specifically includes: performing a differential operation on the first moiré phase image and the second moiré phase image to generate a differential anomaly image, wherein the differential anomaly image is a response difference; The points in the differential anomaly map whose values ​​are higher than a preset threshold are identified as the surface anomaly points, and their position information is output as the measurement results of the surface anomaly points.

6. The method for measuring surface anomalies of a suspension bridge based on an image according to claim 5, characterized in that: The differential operation specifically includes: The first moiré phase image and the second moiré phase image are normalized, and a difference between the normalized first moiré phase image and the second moiré phase image is calculated to generate the differential anomaly image.

7. The method for measuring surface anomalies of a suspension bridge based on an image according to claim 4, characterized in that: The joint solution is specifically: The integrability condition that the two-dimensional tilt field must satisfy is used as an optimization constraint. The first and second moiré phase images are solved by an optimization algorithm to obtain the two-dimensional tilt field. The calculation error caused by optical distortion or component misalignment is corrected based on the two-dimensional tilt field.

8. The image-based suspension bridge tilt measurement method according to claim 1, characterized in that: In step S1 , the first polarization grating and the second polarization grating are both generated and switched using a spatial light modulator.

9. The method for measuring surface anomalies of a suspension bridge based on an image according to claim 1, characterized in that: The first spatial direction is a horizontal direction, and the second spatial direction is a vertical direction.

10. A measurement system for implementing the image-based suspension bridge surface anomaly measurement method according to claim 1, characterized in that: include: a projection unit, configured to project the first polarization grating and the second polarization grating onto the area to be measured of the suspension bridge in a time-sharing manner; a polarization state imaging unit, configured to respectively collect the first polarization information and the second polarization information reflected by the area to be measured; A data processing unit is connected to the projection unit and the polarization state imaging unit, and is used to generate a first moiré phase image and a second moiré phase image based on the first polarization information and the second polarization information, respectively, and process the first moiré phase image and the second moiré phase image to synchronously obtain the tilt measurement results of the measured area and the measurement results of the surface abnormal points.

Images