A method and system for enhancing cavity regions of ground penetrating radar images

Abstract

The application discloses a kind of cavity area enhancement method and system of ground penetrating radar image, belong to city road detection technical field.For solving the problem of realizing stable detection and high-precision edge positioning of cavity.The application adopts ground penetrating radar to collect original image inside road, then carries out standardization processing to obtain standardized image;Frequency domain is converted using Fourier transform, then define spectrum energy and energy gradient to represent frequency variation amplitude, construct frequency difference ratio;Regularization regulation strategy based on first derivative and second derivative fusion is established, the spectrum energy is adjusted to obtain regular smooth energy;Frequency gain function is constructed;Phase disturbance angle model is constructed, then the adjusted phase angle is used, combined with frequency gain function to calculate enhanced spectrum;The reconstruction image of enhanced spectrum and the fusion image of space frequency are fused to obtain the final enhanced image;Local anomaly factor is constructed using the gradient intensity of final enhanced image, for boundary calculation and cavity mask generation.

Description

Technical Field

[0001] This invention belongs to the field of urban road detection technology, specifically relating to a method and system for enhancing the cavity area of ​​ground penetrating radar images. Background Technology

[0002] During the long-term operation of urban road infrastructure, the problem of internal cavities caused by factors such as foundation settlement and aging underground pipelines has become increasingly prominent, gradually evolving into one of the key factors threatening road traffic safety. Traditional manual detection methods or borehole sampling techniques are limited by low efficiency, high cost, and difficulty in large-scale application. To solve these problems, non-contact detection methods, represented by ground-penetrating radar (GPR), have attracted much attention due to their excellent spatial resolution and efficient data acquisition capabilities. This technology achieves non-destructive assessment of the internal condition of a target structure by emitting electromagnetic waves and receiving their reflected signals in the medium. However, in practical applications, due to the influence of multiple factors such as differences in material properties, environmental noise interference, and the complexity of image processing, the two-dimensional or three-dimensional imaging results obtained by GPR are often accompanied by significant artifacts and uncertainties, thus posing a huge challenge to the location and accurate definition of cavities. Such abnormal areas often exhibit characteristic patterns such as uneven gray-scale distribution, smooth edge transitions, or sudden changes in intensity. These subtle changes are still difficult to effectively extract and analyze even with professional image analysis algorithms. Summary of the Invention

[0003] The present invention aims to solve the problem of stable detection and high-precision edge localization of cavities, and proposes a method and system for enhancing the cavity region of ground penetrating radar images.

[0004] To achieve the above objectives, the present invention provides the following technical solution:

[0005] A method for enhancing void regions in ground-penetrating radar images includes the following steps:

[0006] S1. Ground-penetrating radar is used to collect raw images of the road interior, which are then standardized to obtain standardized images;

[0007] S2. The standardized image obtained in step S1 is transformed into the frequency domain using Fourier transform. Then, spectral energy and energy gradient are defined to characterize the frequency change amplitude, and the frequency difference ratio is constructed.

[0008] S3. Establish a regularization control strategy based on the fusion of first and second derivatives to adjust the spectral energy and obtain regularized smooth energy;

[0009] S4. Given that the void reflection structure often exhibits a local high-frequency response in the frequency domain, a frequency gain function is constructed using the energy gradient obtained in step S2 and the regularized smooth energy obtained in step S3.

[0010] S5. Construct a phase perturbation angle model using the frequency difference ratio obtained in step S2, and then calculate the enhanced spectrum using the adjusted phase angle and the frequency gain function;

[0011] S6. The reconstructed image with enhanced spectrum and the spatial frequency fusion map are fused to obtain the final enhanced image;

[0012] S7. The gradient intensity of the final enhanced image is used to construct a local anomaly factor for boundary calculation and hole mask generation.

[0013] Furthermore, the standardization process in step S1 involves normalizing the grayscale of the original image inside the road to obtain a grayscale image, and then performing zero-mean standard deviation unitization to obtain a standardized image.

[0014] Furthermore, the specific implementation method of step S2 includes the following steps:

[0015] S2.1. The standardized image obtained in step S1 is transformed into the image frequency domain using Fourier transform, and the expression is:

[0016]

[0017] in, For the image frequency domain; , These are the indices in the x and y directions of the frequency domain, obtained from the frequency coordinates of the Fourier transform; The imaginary unit, For standardized images;

[0018] S2.2. Define the spectral energy and energy gradient using the image frequency domain, with the following expression:

[0019]

[0020]

[0021] in, For spectral energy; It is the energy gradient mode; The symbol is for partial differentials;

[0022] S2.3. Normalize the energy gradient magnitude and spectral energy to construct the frequency difference ratio, expressed as:

[0023]

[0024] in, This represents the frequency difference ratio.

[0025] Furthermore, step S3 involves constructing a local perturbation regularization factor to adjust the spectral energy and obtain the regularized smoothing energy, expressed as:

[0026]

[0027]

[0028] in, This is the local perturbation regularization factor; This is the regularization factor adjustment coefficient, determined by expert experience, and also used to adjust the dimensions; To obtain the first derivative with respect to the spectral energy; To obtain the second derivative with respect to the spectral energy; It is a regularization weighting factor, determined by expert experience, and is also used to adjust the dimensions; This represents the regularized smoothing energy.

[0029] Furthermore, the expression for the frequency gain function constructed in step S4 is as follows:

[0030]

[0031] in, For frequency gain; This is the maximum response scaling factor, determined by expert experience, and also used to adjust the dimensions; is the exponential control factor, determined by expert experience; tanh is the hyperbolic tangent function.

[0032] Furthermore, the specific implementation method of step S5 includes the following steps:

[0033] S5.1. Construct a phase perturbation angle model using the differential calculation result of the frequency difference ratio along the vertical frequency axis obtained in step S2, with the expression as follows:

[0034]

[0035] in, This refers to the phase perturbation angle; This is the phase perturbation factor, determined by expert experience, and also used to adjust the dimensions;

[0036] S5.2. Calculate the adjusted phase angle using the following formula:

[0037]

[0038] in, The adjusted phase angle; for The phase angle;

[0039] The enhanced spectrum is calculated by combining frequency gain, and the calculation formula is as follows:

[0040]

[0041] in, To enhance the spectrum.

[0042] Furthermore, the specific implementation method of step S6 includes the following steps:

[0043] S6.1. Obtain the reconstructed image by performing an inverse Fourier transform using the enhanced spectrum. ;

[0044] S6.2. Introducing a space-frequency dual control model, the space-frequency response is fused to obtain a space-frequency fusion diagram, the expression of which is:

[0045]

[0046] in, This is the inverse Fourier transform; This is a spatial-frequency fusion diagram; The standardized image adjustment index is determined by expert experience; The gain index is determined by expert experience; To standardize the local brightness standard deviation of an image, a sliding window method can be used for calculation;

[0047] S6.3. Perform weighted fusion of the reconstructed image and the spatial frequency fusion map to obtain the final enhanced image. .

[0048] Furthermore, the specific implementation method of step S7 includes the following steps:

[0049] S7.1. The local anomaly factor is constructed using the gradient intensity of the final enhanced image, expressed as:

[0050]

[0051]

[0052] in, For gradient strength; It is a local anomalous factor; , These are the average and standard deviation of the gradient intensity, respectively, calculated using a local sliding window.

[0053] S7.2. Given a threshold, a hole mask is generated by judging local anomaly factors, expressed as:

[0054]

[0055] in, For the purpose of creating a cavity mask; The threshold for abnormal factors is determined by expert experience.

[0056] A system for enhancing the cavity region of a ground-penetrating radar image includes a processor, a memory, and a computer program stored in the memory and executable on the processor. When the computer program is executed, it implements the steps of the method for enhancing the cavity region of a ground-penetrating radar image.

[0057] The beneficial effects of this invention are:

[0058] The present invention provides a method for enhancing the cavity region of ground penetrating radar images, which mainly solves the problems of weak reflection characteristics, blurred boundaries, and strong interference noise in complex radar images. By accurately extracting the cavity frequency anomaly region and generating a corresponding spatial mask, stable detection and high-precision edge positioning of cavities are achieved.

[0059] This invention discloses a method for enhancing void regions in ground-penetrating radar (GPR) images. It integrates key technologies such as image standardization, spectral transformation, energy gradient calculation, frequency regularization, frequency enhancement, phase perturbation, and space-frequency fusion to form a complete void detection framework. Utilizing frequency domain characteristics, it performs multi-dimensional analysis of the echo signal, focusing on extracting energy distribution features, directional perturbation information, and differences in local abrupt changes, thereby accurately reflecting the signal response differences caused by underground voids. Through multi-level structural gradient extraction and edge segmentation mechanisms, it can effectively locate void contours and generate accurate masks, providing scientific data support for road defect diagnosis.

[0060] The present invention discloses a method for enhancing the cavity region of ground penetrating radar images. It employs multi-level frequency domain enhancement and space-frequency fusion to significantly improve the clarity and prominence of the cavity response area in radar images. Compared with traditional algorithms, this method can suppress the influence of environmental noise, preserve key edge information, and detect the cavity location well even in complex backgrounds. It has good accuracy, robustness, and speed, and has a good effect on the detection of underground cavities in highway maintenance. Attached Figure Description

[0061] Figure 1 This is a flowchart of a method for enhancing the cavity region of a ground-penetrating radar image according to the present invention;

[0062] Figure 2 This is a comparison image of the mask before processing by the method of the present invention;

[0063] Figure 3 This is a mask image processed by the hole region enhancement method for ground-penetrating radar images described in this invention. Detailed Implementation

[0064] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only for explaining the invention and are not intended to limit the invention; that is, the described specific embodiments are merely a part of the embodiments of the invention, and not all of them. The components of the specific embodiments of the invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations, and the invention may also have other embodiments.

[0065] Therefore, the following detailed description of specific embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected specific embodiments of the invention. All other specific embodiments obtained by those skilled in the art based on these specific embodiments without inventive effort are within the scope of protection of this invention.

[0066] To further understand the invention's content, features, and effects, the following specific embodiments are provided, along with accompanying drawings. Figure 1 - Appendix Figure 3 Detailed explanation is as follows:

[0067] Example 1:

[0068] A method for enhancing void regions in ground-penetrating radar images includes the following steps:

[0069] S1. Ground-penetrating radar is used to collect raw images of the road interior, which are then standardized to obtain standardized images;

[0070] Furthermore, in step S1, ground-penetrating radar is used to acquire raw images of the road interior. The standardization process involves normalizing the grayscale of the original image of the road interior to obtain a grayscale image. Then, the image is normalized to zero mean and standard deviation to obtain a standardized image. .

[0071] S2. The standardized image obtained in step S1 is transformed into the frequency domain using Fourier transform. Then, spectral energy and energy gradient are defined to characterize the frequency change amplitude, and the frequency difference ratio is constructed.

[0072] Underground radar imaging data often exhibits significant abrupt changes in grayscale due to media discontinuities or void reflections. These abrupt changes typically correspond to high-frequency components in the frequency domain. To deeply analyze these image features, Fourier transform is needed to convert the original spatial domain signal into a frequency domain representation, revealing its inherent structural information. Within the frequency domain analysis framework, energy accumulation intervals and frequency evolution patterns can be intuitively identified, particularly local high-frequency disturbances caused by voids, whose spectra usually exhibit obvious peaks. Based on this, we propose using spectral energy and its gradient to characterize the amplitude of frequency changes. By constructing a frequency difference ratio and combining the gradient value with energy intensity for normalization, we eliminate regions with high energy but stable changes, thereby focusing on key regions with prominent dynamic characteristics and specific morphologies.

[0073] Furthermore, the specific implementation method of step S2 includes the following steps:

[0074] S2.1. The standardized image obtained in step S1 is transformed into the image frequency domain using Fourier transform, and the expression is:

[0075]

[0076] in, For the image frequency domain; , These are the indices in the x and y directions of the frequency domain, obtained from the frequency coordinates of the Fourier transform; The imaginary unit, For standardized images;

[0077] S2.2. Define the spectral energy and energy gradient using the image frequency domain, with the following expression:

[0078]

[0079]

[0080] in, For spectral energy; It is the energy gradient mode; The symbol is for partial differentials;

[0081] S2.3. Normalize the energy gradient magnitude and spectral energy to construct the frequency difference ratio, expressed as:

[0082]

[0083] in, This represents the frequency difference ratio.

[0084] S3. Establish a regularization control strategy based on the fusion of first and second derivatives to adjust the spectral energy and obtain regularized smooth energy;

[0085] Within the frequency domain analysis framework, ground-penetrating radar (GPR) data acquisition may be affected by environmental noise and discontinuities in boundary sampling, often resulting in numerous non-target-induced abrupt changes in spectral images. These spurious abrupt changes can easily mislead target recognition algorithms in subsequent processing stages, thus affecting the accuracy and reliability of the final results. To address this issue, a regularization control strategy based on the fusion of first and second derivatives is proposed, aiming to specifically suppress extreme fluctuations in the spectral energy distribution. By constructing this regularization factor and applying it to the spectral energy adjustment process, the focus and extraction accuracy of subsequent enhancement algorithms on key information are optimized.

[0086] Furthermore, step S3 involves constructing a local perturbation regularization factor to adjust the spectral energy and obtain the regularized smoothing energy, expressed as:

[0087]

[0088]

[0089] in, This is the local perturbation regularization factor; This is the regularization factor adjustment coefficient, determined by expert experience, and also used to adjust the dimensions; To obtain the first derivative with respect to the spectral energy; To obtain the second derivative with respect to the spectral energy; It is a regularization weighting factor, determined by expert experience, and is also used to adjust the dimensions; This represents the regularized smoothing energy.

[0090] S4. Given that the void reflection structure often exhibits a local high-frequency response in the frequency domain, a frequency gain function is constructed using the energy gradient obtained in step S2 and the regularized smooth energy obtained in step S3.

[0091] Given that void-reflective structures often exhibit local high-frequency responses in the frequency domain, a novel frequency enhancement response function was designed to optimize the characterization of potential high-frequency features. This function compares the frequency gradient difference with the local energy spectrum, normalizes the results, and uses an exponential function to adjust the amplification rate, combined with hyperparameter tuning to achieve fine-tuning of specific frequency bands. This method can significantly enhance the regional characteristics of small frequency jumps caused by voids accompanied by low background energy, thereby improving the model's accuracy and specificity in identifying target structures. To prevent extreme fluctuations in the enhancement effect, a hyperbolic tangent function is introduced to smooth the results, improving the stability of the overall frequency response characteristics and the uniformity of its gradient distribution.

[0092] Furthermore, the expression for the frequency gain function constructed in step S4 is as follows:

[0093]

[0094] in, For frequency gain; This is the maximum response scaling factor, determined by expert experience, and also used to adjust the dimensions; is the exponential control factor, determined by expert experience; tanh is the hyperbolic tangent function.

[0095] S5. Construct a phase perturbation angle model using the frequency difference ratio obtained in step S2, and then calculate the enhanced spectrum using the adjusted phase angle and the frequency gain function;

[0096] Besides spectral amplitude characteristics, phase information also contains important structural orientation and texture attributes. This is because the scattering of electromagnetic waves by the void region not only affects signal strength but also significantly interferes with the phase distribution. Therefore, by rationally designing a phase mapping strategy in the frequency domain, the image's ability to perceive spatial structural changes can be effectively improved. This step uses the differential calculation result of the frequency difference ratio along the vertical frequency axis to construct a perturbation angle model, quantifying the unbalanced characteristics and evolution of the target in different directions in the frequency domain. Through phase adjustment, energy distribution is optimized and controlled during frequency domain reconstruction, making the void edges more prominent and accurate in space.

[0097] Furthermore, the specific implementation method of step S5 includes the following steps:

[0098] S5.1. Construct a phase perturbation angle model using the differential calculation result of the frequency difference ratio along the vertical frequency axis obtained in step S2, with the expression as follows:

[0099]

[0100] in, This refers to the phase perturbation angle; This is the phase perturbation factor, determined by expert experience, and also used to adjust the dimensions;

[0101] S5.2. Calculate the adjusted phase angle using the following formula:

[0102]

[0103] in, The adjusted phase angle; for The phase angle;

[0104] The enhanced spectrum is calculated by combining frequency gain, and the calculation formula is as follows:

[0105]

[0106] in, To enhance the spectrum.

[0107] S6. The reconstructed image with enhanced spectrum and the spatial frequency fusion map are fused to obtain the final enhanced image;

[0108] While frequency domain enhancement helps improve image edges and textures, it can cause structural blurring or loss of detail in low-contrast areas or when heavily influenced by a background. To address this, we first use inverse Fourier transform to obtain the frequency-enhanced image, then add a perceptual enhancement term. This term focuses on areas of significant change based on the image's spatial gradient and local brightness information, taking into account the spatial distribution of the frequency domain gain map to improve the layering of hole areas in the final output. The combination of the two parts is adjusted using a weight, ensuring the edge sensitivity of frequency domain enhancement while incorporating local information from the spatial domain, which helps distinguish holed targets from the background.

[0109] Furthermore, the specific implementation method of step S6 includes the following steps:

[0110] S6.1. Obtain the reconstructed image by performing an inverse Fourier transform using the enhanced spectrum. The expression is:

[0111]

[0112] in, This is the inverse Fourier transform;

[0113] S6.2. Introducing a space-frequency dual control model, the space-frequency response is fused to obtain a space-frequency fusion diagram, the expression of which is:

[0114]

[0115] in, This is the inverse Fourier transform; This is a spatial-frequency fusion diagram; The standardized image adjustment index is determined by expert experience; The gain index is determined by expert experience; To standardize the local brightness standard deviation of an image, a sliding window method can be used for calculation;

[0116] S6.3. Perform weighted fusion of the reconstructed image and the spatial frequency fusion map to obtain the final enhanced image. The expression is:

[0117]

[0118] in, , These are the weighting coefficients for the reconstructed map and the weighting coefficients for the space-frequency fusion map, which are determined by expert experience and are also used to adjust the dimensions.

[0119] S7. The gradient intensity of the final enhanced image is used to construct a local anomaly factor for boundary calculation and hole mask generation.

[0120] To effectively detect holes, an anomaly factor representing the local saliency of the image was constructed using the structure gradient of the final enhanced image. Since holes in ground-penetrating radar images can produce large edge responses in the enhanced image, the gradient magnitude of the entire image was first calculated. Then, the average gradient and standard deviation of the neighborhood of each point were calculated using a sliding window method, thus measuring the contrast saliency between the current point and its background. The greater the contrast saliency, the more likely the edges around this point are an anomalous structures, potentially becoming candidate holes. A threshold was then used to determine if the anomaly factor was large enough. If the anomaly factor was large, its location was added to the mask image, resulting in a binary image representing the hole boundary. One advantage of this approach is that it enhances the edge contrast of the region of interest while statistically removing spurious structure responses.

[0121] Furthermore, the specific implementation method of step S7 includes the following steps:

[0122] S7.1. The local anomaly factor is constructed using the gradient intensity of the final enhanced image, expressed as:

[0123]

[0124]

[0125] in, For gradient strength; It is a local anomalous factor; , These are the average and standard deviation of the gradient intensity, respectively, calculated using a local sliding window.

[0126] S7.2. Given a threshold, a hole mask is generated by judging local anomaly factors, expressed as:

[0127]

[0128] in, For the purpose of creating a cavity mask; The threshold for abnormal factors is determined by expert experience.

[0129] The method proposed in this embodiment yields a clearer structural image and uses a local anomaly factor to obtain a hole mask, thereby achieving better detection of holes and their boundaries in the road, improving the robustness and accuracy of the detection.

[0130] The application examples of this embodiment are illustrated below:

[0131] Processing the ground-penetrating radar image corresponding to a certain cavity area, such as... Figure 2 and Figure 3 The comparison shows that the method of this invention extracts a longer and more complete cavity region, accurately covering the actual area of ​​the cavity. Compared with traditional methods, it avoids region recognition shrinkage, improves the ability to restore the length and boundary of the cavity, and enhances the comprehensiveness and accuracy of the detection.

[0132] Example 2:

[0133] A system for enhancing the cavity region of a ground-penetrating radar image includes a processor, a memory, and a computer program stored in the memory and executable on the processor. When the computer program is executed, it implements the steps of the method for enhancing the cavity region of a ground-penetrating radar image as described in Embodiment 1.

[0134] It should be noted that relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0135] Although this application has been described above with reference to specific embodiments, various modifications can be made and components can be replaced with equivalents without departing from the scope of this application. In particular, as long as there is no structural conflict, the features in the specific embodiments disclosed in this application can be combined with each other in any way. The lack of an exhaustive description of these combinations in this specification is merely for the sake of brevity and resource conservation. Therefore, this application is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. A method for enhancing the cavity region of a ground-penetrating radar image, characterized in that, Includes the following steps: S1. Ground-penetrating radar is used to collect raw images of the road interior, which are then standardized to obtain standardized images; S2. The standardized image obtained in step S1 is transformed into the frequency domain using Fourier transform. Then, spectral energy and energy gradient are defined to characterize the frequency change amplitude, and the frequency difference ratio is constructed. The specific implementation method of step S2 includes the following steps: S2.

1. The standardized image obtained in step S1 is transformed into the image frequency domain using Fourier transform, and the expression is: ； in, For the image frequency domain; , These are the indices in the x and y directions of the frequency domain, obtained from the frequency coordinates of the Fourier transform; The imaginary unit, For standardized images; S2.

2. Define the spectral energy and energy gradient using the image frequency domain, with the following expression: ； ； in, For spectral energy; It is the energy gradient mode; The symbol is for partial differentials; S2.

3. Normalize the energy gradient magnitude and spectral energy to construct the frequency difference ratio, expressed as: ； in, Frequency difference ratio; S3. Establish a regularization control strategy based on the fusion of first and second derivatives to adjust the spectral energy and obtain regularized smooth energy; Step S3 involves constructing a local perturbation regularization factor and adjusting the spectral energy to obtain the regularized smoothing energy, expressed as: ； ； in, This is the local perturbation regularization factor; This is the regularization factor adjustment coefficient, determined by expert experience, and also used to adjust the dimensions; To obtain the first derivative with respect to the spectral energy; To obtain the second derivative with respect to the spectral energy; It is a regularization weighting factor, determined by expert experience, and is also used to adjust the dimensions; For regularized smoothing energy; S4. Given that the void reflection structure often exhibits a local high-frequency response in the frequency domain, a frequency gain function is constructed using the energy gradient obtained in step S2 and the regularized smooth energy obtained in step S3. The expression for the frequency gain function constructed in step S4 is: ； in, For frequency gain; This is the maximum response scaling factor, determined by expert experience, and also used to adjust the dimensions; The exponential control factor is determined by expert experience; tanh is the hyperbolic tangent function. S5. Construct a phase perturbation angle model using the frequency difference ratio obtained in step S2, and then calculate the enhanced spectrum using the adjusted phase angle and the frequency gain function; The specific implementation method of step S5 includes the following steps: S5.

1. Construct a phase perturbation angle model using the differential calculation result of the frequency difference ratio along the vertical frequency axis obtained in step S2, with the expression as follows: ； in, This refers to the phase perturbation angle; This is the phase perturbation factor, determined by expert experience, and also used to adjust the dimensions; S5.

2. Calculate the adjusted phase angle using the following formula: ； in, The adjusted phase angle; for The phase angle; The enhanced spectrum is calculated by combining frequency gain, and the calculation formula is as follows: ； in, To enhance the spectrum; S6. The reconstructed image with enhanced spectrum and the spatial frequency fusion map are fused to obtain the final enhanced image; S7. The gradient intensity of the final enhanced image is used to construct a local anomaly factor for boundary calculation and hole mask generation.

2. The method for enhancing the cavity region of a ground-penetrating radar image according to claim 1, characterized in that, The standardization process in step S1 involves normalizing the grayscale of the original image inside the road to obtain a grayscale image, and then performing zero-mean standard deviation unitization to obtain a standardized image.

3. The method for enhancing the cavity region of a ground-penetrating radar image according to claim 2, characterized in that, The specific implementation method of step S6 includes the following steps: S6.

1. Obtain the reconstructed image by performing an inverse Fourier transform using the enhanced spectrum. ; S6.

2. Introducing a space-frequency dual control model, the space-frequency response is fused to obtain a space-frequency fusion diagram, the expression of which is: ； in, This is the inverse Fourier transform; This is a spatial-frequency fusion diagram; The standardized image adjustment index is determined by expert experience; The gain index is determined by expert experience; To standardize the local brightness standard deviation of the image, a sliding window method was used for calculation; S6.

3. Perform weighted fusion of the reconstructed image and the spatial frequency fusion map to obtain the final enhanced image. .

4. The method for enhancing the cavity region of a ground-penetrating radar image according to claim 3, characterized in that, The specific implementation method of step S7 includes the following steps: S7.

1. The local anomaly factor is constructed using the gradient intensity of the final enhanced image, expressed as: ； ； in, For gradient strength; It is a local anomalous factor; , These are the average and standard deviation of the gradient intensity, respectively, calculated using a local sliding window. S7.

2. Given a threshold, a hole mask is generated by judging local anomaly factors, expressed as: ； in, For the purpose of creating a cavity mask; The threshold for abnormal factors is determined by expert experience.

5. A system for enhancing the cavity region of a ground-penetrating radar image, characterized in that, It includes a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed, implements the steps of a method for enhancing the cavity region of a ground-penetrating radar image as described in any one of claims 1-4.

Images