Three-dimensional ground penetrating radar imaging method based on interface constraint coherent weighting and radar

Abstract

The present application relates to the field of road nondestructive testing, and discloses a three-dimensional ground penetrating radar imaging method based on interface constraint coherent weighting and a radar. The method comprises the following steps: discretizing a road section to be detected into a three-dimensional voxel grid, and calculating the two-way travel time of electromagnetic waves from the ground surface to the grid points based on the electromagnetic wave propagation speed. The signal amplitude of the radar data is extracted based on the two-way travel time, and the stacking energy and the coherence factor of the grid points are calculated. The interface penalty weight of the sudden interface to the grid points is calculated and weighted with the stacking energy and the coherence factor. After traversing all the grid points, the ground penetrating radar data volume is generated. Through the deep integration of the continuous steps of first arrival wave correction, layered velocity modeling, weighted inverse projection and customized physical penalty weight, the divergent hyperbolic curve characteristics in the original radar image can be converged and focused into the real disease geometry, the lateral resolution is significantly improved, the disease boundary can be accurately identified, and the depth positioning error is reduced.

Description

Technical Field

[0001] This invention relates to the field of road non-destructive testing technology, and in particular to a three-dimensional ground-penetrating radar imaging method and radar based on interface-constrained coherent weighting. Background Technology

[0002] Three-dimensional ground-penetrating radar (3D GPR) is a non-destructive testing technology that uses high-frequency electromagnetic waves to probe underground media. In highway maintenance, 3D GPR is widely used to detect hidden defects such as cavities, voids, and looseness beneath the pavement. Existing commercial GPR systems primarily use a "time-depth conversion" method based on a single velocity model to generate underground profiles. This assumes the underground medium is homogeneous and uses formulas to convert the two-way travel time of radar waves into depth. However, this existing technology has the following drawbacks: 1. Blurred imaging, unable to reconstruct the true morphology of the damage: Radar antennas have a certain beamwidth. When the radar passes over a point target (such as an underground cavity), echoes can be received both directly above and within a certain distance in front of and behind the target. This causes the actual point target to appear as a diverging hyperbola or parabola in the radar image. Existing simple conversions cannot converge this divergent energy, resulting in blurred boundary lines of the damage in the image, severely affecting the estimation of the damage volume. 2. Large depth positioning error: Highway pavement media are complex, typically composed of an upper layer, middle layer, lower layer (asphalt concrete), and base layer (water-stabilized crushed stone), with significant differences in dielectric constant between layers. Existing technologies often use a single average velocity, ignoring interlayer refraction effects, leading to positioning errors of over 10% for deep damage. 3. Low signal-to-noise ratio, numerous artifacts: The underground environment is complex, containing a large amount of random noise and interlayer multiple wave interference. Traditional imaging algorithms lack effective detection mechanisms and cannot distinguish between valid signals and sidelobe noise. 4. The dielectric constant at the interface between different media layers will change abruptly. For example, at the interface between the asphalt lower layer and the water-stabilized base layer, extremely strong and highly consistent physical reflections will occur in multiple channels. This will cause the polarization failure of the conventional CF (coherence factor) algorithm to abnormally approach 1 at this depth, generating a horizontally bright false artifact that runs through the 3D image. This severely interferes with the identification of hidden defects at this depth, thus masking the real defects. Summary of the Invention

[0003] To address the technical problems of blurred imaging and inaccurate depth positioning of existing ground-penetrating radars, this invention provides a three-dimensional ground-penetrating radar imaging method and radar based on interface-constrained coherent weighting.

[0004] In a first aspect, this invention proposes a three-dimensional ground-penetrating radar imaging method based on interface-constrained coherent weighting, comprising: Obtain the road segment to be detected K Material properties and thickness of the layer mediumh k Calculation based on material properties k electromagnetic wave propagation speed of layered medium v k . k ∈[1, K The interface between adjacent media constitutes... M A mutation interface. Based on h k Calculate the first m Depth of each mutation interface z inter,m .

[0005] The road segment to be detected is discretized into a three-dimensional voxel mesh, and based on... K Calculation of electromagnetic wave propagation speed in a layered medium. Electromagnetic waves travel from the Earth's surface to the third layer in a three-dimensional voxel grid. p grid points P ( x p , y p , z p Equivalent speed v rms ( z p ).

[0006] according to v rms ( z p ) Calculate the first in ground-penetrating radar i One antenna to P ( x p , y p , z p Two-way travel time t i,p and extract the first i Each antenna at time t i,p signal amplitude at D i ( t i,p Then iterate through I One antenna, based on D i ( t i,p )calculate P ( x p , y p , z pThe superposition energy of ) S ( P and coherence factor CF ( P ). i ∈[1 , I ].

[0007] calculate P ( x p , y p , z p Interface penalty weight for strong interlayer reflection caused by abrupt changes in dielectric constant at the interface. W ( z p ): .

[0008] α m For the first m The penalty coefficient for each mutation interface. c m For the first m The influence range constant of each mutation interface. S ( P ), CF ( P ), W ( z p After fusion, as P ( x p , y p , z p The pixel values ​​of the grid points are used to generate the ground-penetrating radar data volume after traversing all grid points.

[0009] Secondly, the present invention also proposes a three-dimensional ground-penetrating radar based on interface constraint coherent weighting, which uses the three-dimensional ground-penetrating radar imaging method based on interface constraint coherent weighting in the first aspect for non-destructive testing of underground road structures.

[0010] Thirdly, the present invention also proposes a computer-readable storage medium having a computer program / instructions stored thereon. When executed by a processor, the computer program / instructions implement the steps of the interface-constrained coherent weighting-based three-dimensional ground-penetrating radar imaging method in the first aspect.

[0011] Fourthly, the present invention also proposes a computer program product, comprising a computer program / instructions. When executed by a processor, the computer program / instructions implement the steps of the interface-constrained coherent weighting three-dimensional ground-penetrating radar imaging method in the first aspect.

[0012] The beneficial effects of this invention are as follows: This invention employs a deep fusion process involving consecutive steps such as first-arrival correction, layered velocity modeling, weighted inverse projection, and customized physical penalty weights. This process converges and focuses the divergent hyperbolic features (synthetic aperture length) in the original radar image into the true geometry of the defects, significantly improving lateral resolution and facilitating accurate identification of defect boundaries. Simultaneously, layered velocity modeling effectively reduces depth positioning errors for deep defects (such as voids at the bottom of the base layer), providing accurate depth information for maintenance grouting operations. The introduced coherence factor weighting technique adaptively suppresses incoherent sidelobe noise and pavement multiple wave interference, resulting in a cleaner image background and highlighting the characteristics of hidden structural defects. Furthermore, by introducing interface penalty weights, it overcomes the specific inter-layer interface in-phase interference caused by typical multi-layer pavement structures on highways, ultimately leading to clear ground-penetrating radar imaging and accurate defect depth positioning. Attached Figure Description

[0013] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0014] Figure 1 This is a flowchart of a three-dimensional ground-penetrating radar imaging method based on interface-constrained coherent weighting; Figure 2 This is a schematic diagram illustrating the principle of first-arrival alignment based on envelope statistics. Figure 3 This is a schematic diagram illustrating the process of modeling the velocity field of a non-uniform medium layer. Figure 4 This is a schematic diagram illustrating the principle of coherence factor-weighted inverse projection imaging. Detailed Implementation

[0015] The technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0016] 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 invention pertains. The terminology used herein in the specification of this invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "or / and" as used herein includes any and all combinations of one or more of the associated listed items.

[0017] Please refer to Figure 1 This embodiment provides a three-dimensional ground-penetrating radar imaging method based on interface-constrained coherent weighting, which mainly includes deep fusion of four core aspects: first-arrival alignment based on envelope statistics, layered velocity field modeling in non-uniform media, coherence factor-weighted back projection (BP), and a Gaussian concave penalty weighting function with specific correction. The imaging method and core aspects are described in detail below.

[0018] I. First arrival wave alignment based on envelope statistics.

[0019] Due to the vibrations caused by the vehicle's movement and hardware-triggered jitter, the radar data collected by different antenna channels of the ground-penetrating radar exhibits a slight deviation at "time zero (the moment of surface reflection)," which directly affects the accuracy of subsequent imaging. This step is equivalent to processing the raw radar data obtained by the ground-penetrating radar. x ( t Preprocessing is performed to time-align the data before generating the ground-penetrating radar data volume, in order to ensure consistency. x ( t The surface reflection time of ( ). For example, Figure 2 As shown in this embodiment, the time alignment process includes: First, read the raw A-scan data from the ground-penetrating radar. Then, use Hilbert transform to extract the raw radar data. x ( t Instantaneous envelope E ( t ): .

[0020] In the formula, H [·] denotes the Hilbert transform operator. Furthermore, an adaptive threshold is set. T th : T th = q · σ .in, q The empirical coefficient (which can be taken as follows in this embodiment) q =4). σ Raw radar data x ( t The standard deviation of the background noise. Then, a search from scratch is performed on the timeline.E ( t First time exceeding T th The moment t 0. Then take the raw radar data. x ( t Shift the time on the time axis t 0 is equivalent to truncating the time. t Data prior to 0, making each raw radar data... x ( t The surface reflection time of each antenna channel is uniformly assigned to 0. This step, through Hilbert transform and thresholding, eliminates time jitter during hardware acquisition, ensuring that the data from all antenna channels are aligned on the time axis, which is a prerequisite for subsequent multi-channel overlay.

[0021] II. Modeling of velocity fields in non-uniform media.

[0022] To correct for errors caused by different layered media in roads, this embodiment constructs a layered velocity model that conforms to the actual road structure, such as... Figure 3 As shown, its construction method includes: First, obtain the road design drawings of the section to be tested, and determine the thickness of each layer of medium using the structural design parameters and calibration data on them. h k For example, from top to bottom, it includes: the upper layer h 1: h 1=0.04m, intermediate layer h 2: h 2=0.06m, lower layer h 3: h 3 = 0.08m, etc., from which we can deduce all K The material properties, dielectric constant, thickness, and other parameters of the media are crucial. If road design drawings are unavailable for various reasons, these parameters can be obtained through core sampling and calibration. Once these parameters are known, the electromagnetic wave propagation speed of each media layer can be calculated. v k : .

[0023] In the formula, c This represents the speed of light in a vacuum (the speed of light can be approximated as 0.3 m / ns). ε r,k Indicates the first k The relative permittivity of the dielectric layer. k ∈[1, KThrough the above calculations, the electromagnetic wave propagation speed of each medium layer is known, allowing the construction of a discrete velocity field function that varies with depth. This step, tailored to the unique layered structure of roads, no longer uses a single velocity but establishes a velocity field based on physical layering. This enables the calculation of the "time of flight" of ground-penetrating radar waves to take into account the refraction effects of different medium layers, thereby correcting for nonlinear errors related to depth.

[0024] III. Back projection of coherence factor weighted (BP).

[0025] Based on the above preprocessed raw radar data x ( t The system processes the energy using the principles of Synthetic Aperture Radar (SAR), repositioning the divergent energy through "back projection" and introducing a "coherence factor" to suppress noise. For example... Figure 4 As shown, in this embodiment, the coherence factor-weighted inverse projection imaging method includes: First, the road segment to be inspected is meshed in three dimensions, then discretized into three-dimensional voxel meshes of equal size, and the resolution of the three-dimensional voxel mesh is set as follows: Δ x Δ y Δ z = 0.02m. Δ x Δ y Δ z These represent the length, width, and height of one of the grid cells. If a total of [number] grid cells are divided... P For the 3D voxel mesh, then for the 3D voxel mesh p For a three-dimensional voxel mesh, its three-dimensional center is defined as a mesh point of that three-dimensional voxel mesh. P ( x p , y p , z p ). p ∈[1, P In subsequent data processing, grid points were used as the basis. P ( x p , y p , z p ) are used as coordinates for calculation.

[0026] Subsequently, ray tracing and travel time calculations were performed on the ground-penetrating radar's antenna and three-dimensional voxel mesh. This ground-penetrating radar includes... I There are 1 antenna, and the position of each antenna can be recorded as: A i ( x i , y i , 0).i ∈[1, I First, calculate the electromagnetic wave's trajectory from the Earth's surface to the 3D voxel mesh. p grid points P ( x p , y p , z p Equivalent speed v rms ( z p ): ; ; in, x p , y p , z p These represent the x-coordinate, y-coordinate, and depth coordinates of the underground grid points, i.e., the three-dimensional spatial coordinates. Δ t k Indicates the electromagnetic wave at the th k The vertical one-way propagation time within the medium layer. Using this root-mean-square velocity approximation algorithm to process the electromagnetic wave refraction path can improve computational efficiency while ensuring the accuracy of the calculation results. This is based on the electromagnetic wave's arrival time at the [missing information - likely a specific point or stage]. p grid points P ( x p , y p , z p Equivalent speed v rms ( z p When ), the third digit in the three-dimensional ground-penetrating radar can be calculated. i The electromagnetic waves emitted by each antenna to P ( x p , y p , z p Two-way travel time t i,p (i.e., the total time of electromagnetic wave incident-reflection-reception): .

[0027] x i , y i They represent the first iThe antenna's horizontal and vertical coordinates on the Earth's surface, i.e., its two-dimensional horizontal coordinates. Due to the aforementioned first-arrival alignment steps, the consumed two-way travel time... t i,p All statistics are calculated starting from time 0, ensuring the two-way travel time is accurate. t i,p This also represents starting from time 0 on the timeline. t i,p At this moment, the first [time] is extracted from the radar data. i Each antenna at time t i,p signal amplitude at D i ( t i,p For example, the two-way travel time of an electromagnetic wave emitted by an antenna at time 0 to a grid point. t i,p Since the time interval is 40ns, the subsequent signal amplitude needs to be extracted at 40ns on the time axis. Following the above logic, iterate... I One antenna, based on D i ( t i,p )calculate P ( x p , y p , z p The superposition energy of ) S ( P and coherence factor CF ( P In this embodiment, superimposed energy S ( P The formula for calculating ) is: .

[0028] coherence factor CF ( P The formula for calculating ) is: .

[0029] This coherence factor CF ( P The value range of ) is 0 to 1. When the signals of each antenna channel are in phase, CF ( P The value approaches 1. When the signal is mainly random noise, CF ( P The value approaches 0.

[0030] IV. Gaussian concave penalty weight function with specific modifications.

[0031] In actual highway testing, the dielectric constant at the interface between different media can change abruptly. For example, the dielectric constant at the interface between the asphalt surface layer and the water-stabilized base layer can change from 6.0 to 8.0, which can severely interfere with the identification of latent defects at that depth. To address this, an innovative interface penalty weight, varying with depth, is constructed and introduced to account for the strong interlayer reflection caused by the abrupt change in dielectric constant. W ( z p This allows for adaptive suppression of multiple mutation interfaces. ; Where Π is the multiplication symbol. The boundary between adjacent media constitutes... M A mutation interface. z inter,m For the first m The depth of a mutation interface can be determined by the thickness of the medium. h k Calculated. α m For the first m The penalty coefficient for abrupt interface is positively correlated with the difference in relative permittivity between the two adjacent media; the greater the difference, the larger the value. The value ranges from 0.55 to 0.95, and 0.8 can be used for the interface between a typical asphalt surface layer and a semi-rigid water-stabilized base course. c m For the first m The influence range constant of the abrupt interface is used to control the vertical thickness of the penalty suppression zone (i.e., the standard deviation of the Gaussian function). It is positively correlated with the interface depth and has a value range of 0.03m to 0.1m. In this example, it can be taken as 0.06m to strictly match the spatial pulse width of the electromagnetic wavelet of high-frequency ground penetrating radar in the corresponding medium.

[0032] In the final image output, energy will be superimposed. S ( P coherence factor CF ( P Interface penalty weight W ( z p After multiplication, it becomes P ( x p , y p , z p ) pixel values I final ( P ): I final ( P )= S ( P )× CF( P )× W ( z p The weighted pixel value I final ( P As the first p grid points P ( x p , y p , z p The final pixel value of the signal is obtained by traversing all grid points to generate a complete ground-penetrating radar data volume. Compared to traditional radar maps that display "echo arrival time," this invention calculates the distance from each grid point to all antennas, "backtracking" the received signal back to its actual underground spatial location, avoiding the drawbacks of traditional radar imaging. Simultaneously, this invention utilizes the phase consistency of multi-channel signals. Real defects (such as cavities) will produce phase-consistent echoes on all antennas, while noise is random. This is achieved by multiplying by a coherence factor. CF ( P This automatically preserves the true signal and suppresses background clutter, thus achieving high signal-to-noise ratio imaging. Through the aforementioned specific calculations, when the grid points... P ( x p , y p , z p When approaching any mutation interface, the corresponding item decreases, and the overall interface penalty weight decreases. W ( z p The Gaussian concavity decreases sharply, adaptively penalizing and suppressing interface artifacts caused by abrupt changes in the medium. When the mesh points... P ( x p , y p , z p When the medium is inside the medium layer and far from all abrupt interfaces, all Gaussian terms approach 0 and recover to approach 1, which has no effect on the imaging of latent defects. Thus, while preserving the real in-phase defect signal, it accurately suppresses the horizontal artifact blind zone generated by random background clutter and abrupt interfaces of specific dielectric constants of highways from the root of the algorithm.

[0033] To verify the effectiveness of this invention, comparative analysis and effect verification were conducted based on real highway engineering parameters. The process and results are as follows: 1. Simulation environment and road structure parameter settings.

[0034] A typical three-dimensional highway structure model is constructed, with the specific parameters set as follows: (1) Pavement Layering Model: Primarily considering the surface layer and base layer. High-grade highway pavement comprises three main media layers. The first layer is the top layer, 0.04m thick, with a relative permittivity of 5; the second layer is the intermediate layer, 0.06m thick, with a relative permittivity of 5.5; the third layer is the bottom layer, 0.08m thick, with a relative permittivity of 6. The base layer is 0.36m thick, with a relative permittivity of 8.

[0035] (2) Target disease model: A spherical cavity with a diameter of 0.10m (relative permittivity of 1) is set as the target of the hidden disease at a depth of 0.30m inside the pavement structure (i.e. inside the base layer).

[0036] (3) Noise and interference settings: Gaussian white noise is introduced into the simulation background, and multiple wave interference generated by interlayer reflection is simulated to highly restore the real complex road surface detection environment.

[0037] 2. Comparative analysis and effect verification.

[0038] The "traditional processing method (using a single velocity model and conventional unweighted imaging)" that does not employ the core steps of this invention is used as a control group. The results are compared with the 3D data volume processed by the method of this invention (including inverse projection imaging with non-uniform medium layered velocity field modeling, coherence factor weighting, and specific correction), as follows: Verification 1: Improved accuracy in locating disease depth.

[0039] (1) Control group performance: The traditional method uses a single empirical average speed to perform a simple time-depth conversion on the two-way travel time. For diseases with a true depth of 0.30m, the theoretical positioning depth will drift downwards significantly, and the depth positioning error can usually reach more than 10%.

[0040] (2) The invention features: The invention constructs a non-uniform medium layered velocity field and introduces the root mean square equivalent velocity to calculate and compensate for the multi-layer refraction path. In principle, this step takes into account the actual travel time of electromagnetic waves in different layers, and theoretically reduces the nonlinear positioning deviation caused by interlayer refraction of deep defects to a minimum, approaching the actual physical depth of 0.30m.

[0041] Verification 2: Focus on spatial resolution and disease morphology.

[0042] (1) Control group performance: Due to the beam diffusion effect of the three-dimensional ground-penetrating radar antenna, on the original radar profile without reverse projection processing, point targets usually present a divergent hyperbola shape with a span of more than 0.10m in spatial geometry, which cannot reflect the true situation of the disease.

[0043] (2) Performance of the invention: The present invention employs a reverse projection algorithm, which uses ray tracing and travel time calculation to back-project the divergent electromagnetic wave energy along a hyperbolic trajectory and superimpose it onto the actual underground spatial location. Under the geometric convergence limit, the divergent energy achieves high convergence, and the lesion morphology can theoretically be focused back to a high-energy region close to its actual diameter, significantly improving the resolution.

[0044] Verification 3: Improvement in anti-interference capability and signal-to-noise ratio.

[0045] (1) Control group performance: Conventional imaging algorithms lack an effective coherence detection mechanism, the background is filled with random noise, the target signal is easily submerged by the sidelobe energy during the superposition process, and the overall signal-to-noise ratio is limited.

[0046] (2) Invention Features: This invention introduces a coherence factor for adaptive weighting during the superimposed imaging process. From a mathematical perspective: In the target disease area, because the phases of the multi-channel signals are completely consistent, the numerator of the coherence factor is maximized, and its value approaches 1, thus preserving the true signal energy completely; in the random noise area without disease, due to the disordered signal phases, the terms cancel each other out, causing the numerator to approach 0, and the clutter energy is greatly weakened by multiplicative cascading. Theoretically, adaptive suppression of sidelobe noise is achieved, significantly enhancing anti-interference capability.

[0047] Verification 4: The effect of suppressing false artifacts at the mutation interface.

[0048] (1) Control group performance (using only conventional CF-weighted BP imaging): According to the aforementioned pavement layering model, at the junction of the asphalt lower layer and the base layer, i.e., at an absolute depth z = 0.04m + 0.06m + 0.08m = 0.18m, a strong physical surface reflection is generated due to the sudden change in dielectric constant from 6 to 8. The phase of the reflected echoes received by the multi-channel is highly consistent, causing the polarization constraint of the conventional CF algorithm to completely fail at this depth (the CF value approaches 1 across the entire frequency band). In the final output three-dimensional image, a bright horizontal artifact (strong reflective bright band) is generated at a depth of 0.18m that runs through the entire profile. As a result, the target signal of the spherical cavity originally set at a depth of 0.30m is severely interfered with or even obscured, which can easily lead to missed detection in engineering.

[0049] (2) Performance of the Invention: The algorithm of this invention automatically calculates and extracts the strong abrupt change interface = 0.18m as the core spatial target point based on the input design thickness of each surface layer, and matches the corresponding penalty coefficient. During the inverse projection calculation, when the depth of the imaging grid point is close to 0.18m, the penalty weight... W ( z pThe Gaussian decay term rapidly decreases to 0.2, suppressing strong surface reflections of the same phase. However, as the grid points continue to extend downwards, reaching a depth of 0.30m where the target defect is located, the Gaussian decay term approaches 0 due to the distance from the abrupt change interface. W ( z p The image was restored to 1. The final imaging comparison results showed that the high-brightness horizontal artifact at 0.18m was significantly suppressed by the image; due to the complete elimination of strong reflection interference, the spherical cavity at 0.30m was clearly displayed with an extremely high signal-to-noise ratio.

[0050] In another embodiment, a ground-penetrating radar based on interface constraint coherent weighting is proposed, which uses the ground-penetrating radar data volume imaging method based on interface constraint coherent weighting as described in the above embodiment for non-destructive testing of underground road structures.

[0051] In another embodiment, a computer-readable storage medium storing a computer program is also proposed. When the computer program is executed by a processor, it implements the steps of the interface-constrained coherent weighted ground-penetrating radar data volume imaging method described in the above embodiments. The computer-readable storage medium may include, but is not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof.

[0052] In another embodiment, a computer program product is also proposed, comprising a computer program / instructions. This computer program / instructions are used to cause a computer to perform the steps of the interface-constrained coherent weighted ground-penetrating radar data volume imaging method described in the above embodiments. The computer program instructions exist in a computer-readable medium in forms including, but not limited to, source files, executable files, and installation package files. Correspondingly, the computer program instructions can be executed by a computer in ways including, but not limited to: the computer directly executing the instructions; the computer compiling the instructions and then executing the corresponding compiled program; the computer reading and executing the instructions; or the computer reading and installing the instructions and then executing the corresponding installed program.

[0053] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0054] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this invention patent should be determined by the appended claims.

Claims

1. A three-dimensional ground-penetrating radar imaging method based on interface-constrained coherent weighting, characterized in that, include: Obtain the road segment to be detected K Material properties and thickness of the layer medium h k ;Calculation based on material properties k electromagnetic wave propagation speed of layered medium v k ; k ∈[1, K The interface between adjacent media constitutes a total of 1,000 units. M A mutation interface; based on h k Calculate the first m Depth of each mutation interface z inter,m ; The road segment to be detected is discretized into a three-dimensional voxel mesh, and based on... K Calculation of electromagnetic wave propagation speed in a layered medium. Electromagnetic waves travel from the Earth's surface to the third layer in a three-dimensional voxel grid. p grid points P ( x p , y p , z p Equivalent speed v rms ( z p ); according to v rms ( z p ) Calculate the first in ground-penetrating radar i One antenna to P ( x p , y p , z p Two-way travel time t i,p and extract the first i Each antenna at time t i,p signal amplitude at D i ( t i,p ); Subsequently Traversal I One antenna, based on D i ( t i,p )calculate P ( x p , y p , z p The superposition energy of ) S ( P and coherence factor CF ( P ); i ∈[1 , I ]; x p , y p , z p They represent the first p The x-coordinate, y-coordinate, and depth coordinates of each grid point; calculate P ( x p , y p , z p Interface penalty weight for strong interlayer reflection caused by abrupt changes in dielectric constant at the interface. W ( z p ): ； α m For the first m The penalty coefficient for each mutation interface. c m For the first m The influence range constant of each mutation interface; Will S ( P ), CF ( P ), W ( z p After fusion, as P ( x p , y p , z p The pixel values ​​of the grid points are used to generate the ground-penetrating radar data volume after traversing all grid points.

2. The three-dimensional ground-penetrating radar imaging method based on interface-constrained coherent weighting according to claim 1, characterized in that, raw radar data x ( t Time alignment is performed before generating the ground-penetrating radar data volume to ensure consistency. x ( t The surface reflection time.

3. The three-dimensional ground-penetrating radar imaging method based on interface-constrained coherent weighting according to claim 2, characterized in that, Time alignment methods include: Extracting using Hilbert transform x ( t Instantaneous envelope E ( t ); Set adaptive threshold T th : T th = q · σ ;in, q This is an empirical coefficient. σ for x ( t The standard deviation of background noise; Record E ( t First time exceeding T th The moment t 0; Will x ( t Translate on the timeline t 0, making x ( t The surface reflection time of ) is uniformly corresponding to 0.

4. The three-dimensional ground-penetrating radar imaging method based on interface-constrained coherent weighting according to claim 1, characterized in that, In the section of road to be inspected K Material properties and thickness of the layer medium h k Obtained by consulting road design drawings; or by obtaining calibration through core sampling.

5. The three-dimensional ground-penetrating radar imaging method based on interface-constrained coherent weighting according to claim 1, characterized in that, The resolution of the 3D voxel mesh is set to: Δ x Δ y Δ z All are set to 0.02m; Δ x Δ y Δ z These represent the length, width, and height of one of the grid cells, respectively.

6. The three-dimensional ground-penetrating radar imaging method based on interface-constrained coherent weighting according to claim 1, characterized in that, equivalent speed v rms ( z p The formula for calculating ) is: ； ； in, x p , y p , z p Δ represents the x-coordinate, y-coordinate, and depth coordinates of the underground grid points, respectively. t k This indicates that electromagnetic waves are present in the underground structure of the road section. k Vertical one-way propagation time within the layered medium; And / or, two-way travel time t i,p The calculation formula is: ； In the formula, ( x i , y i ) indicates the first i The horizontal coordinates of the antenna on the ground.

7. The three-dimensional ground-penetrating radar imaging method based on interface-constrained coherent weighting according to claim 1, characterized in that, Superimposed energy S ( P The formula for calculating ) is: ； coherence factor CF ( P The formula for calculating ) is: 。 8. A three-dimensional ground-penetrating radar based on interface-constrained coherent weighting, characterized in that, It uses the interface-constrained coherent weighted three-dimensional ground-penetrating radar imaging method as described in any one of claims 1 to 7 to perform non-destructive testing of underground road structures.

9. A computer-readable storage medium having a computer program / instructions stored thereon, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the three-dimensional ground-penetrating radar imaging method based on interface constraint coherent weighting as described in any one of claims 1 to 7.

10. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the three-dimensional ground-penetrating radar imaging method based on interface constraint coherent weighting as described in any one of claims 1 to 7.

Images