Underground pipeline image intelligent noise reduction processing system based on ground penetrating radar

Abstract

The invention discloses an underground pipeline image intelligent noise reduction processing system based on a ground penetrating radar, and belongs to the technical field of data processing, and the system comprises a ground penetrating radar image collection module which is used for collecting ground penetrating radar images of infrastructures of a preset underground pipeline, and constructing an image database based on the collected ground penetrating radar images; and the image data processing module responds to the image database and is used for processing the acquired ground penetrating radar image, including performing image format conversion and / or size adjustment on the ground penetrating radar image. By means of noise detection and noise reduction processing, intelligent algorithm selection and optimization, complex structure processing and the like, the noise reduction effect and the image quality of the ground penetrating radar image are improved, and more accurate and more reliable image data support is provided for detection and maintenance of underground pipelines.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to an intelligent noise reduction processing system for underground pipeline images based on ground penetrating radar. Background Art

[0002] Ground penetrating radar (GPR) is an instrument that uses high-frequency electromagnetic waves for underground exploration. It can be used to survey infrastructure such as airport runways and drainage systems. In practice, GPR images are often affected by various noise factors, such as environmental noise and equipment noise. This noise degrades image quality and affects the accurate detection and assessment of underground pipeline structures. Therefore, noise suppression is crucial.

[0003] Regarding this technical research, application number CN201710656970.0 provides a method and system for denoising ground-penetrating radar (GPR) B-scan images based on EEMD and permutation entropy. This technical solution first acquires a two-dimensional GPR B-scan image signal, then denoises each acquired B-scan image signal to obtain a noise-suppressed signal for that B-scan image signal. Finally, the denoised signals are recombined to obtain a noise-suppressed two-dimensional B-scan image signal. This technical solution addresses the signal modal aliasing problem present in EMD decomposition and can reduce noise to a certain extent.

[0004] However, the above-mentioned technical solutions still have limitations in actual use. For example, although the empirical mode decomposition (EMD) method is adaptive, it has problems of modal aliasing and dependence on experience, making it difficult to accurately grasp the algorithm adjustments for different signal characteristics and / or shapes with complex structures. As a result, the image will be over-smoothed while removing noise, resulting in the loss of useful image details and affecting the accurate detection of the internal structure of underground pipelines. Summary of the Invention

[0005] In view of the above problems existing in the existing data processing technology field, the present invention is proposed.

[0006] Therefore, one of the objects of the present invention is to provide an intelligent noise reduction processing system for underground pipeline images based on ground penetrating radar, which improves the noise reduction effect and image quality of ground penetrating radar images through noise detection and noise reduction processing, intelligent algorithm selection and optimization, complex structure processing and other means, and provides more accurate and reliable image data support for the detection and maintenance of underground pipelines.

[0007] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0008] The present invention provides an intelligent noise reduction processing system for underground pipeline images based on ground penetrating radar, comprising:

[0009] A ground penetrating radar image acquisition module, used to acquire ground penetrating radar images of the infrastructure of the preset underground pipeline and to construct an image database based on the acquired ground penetrating radar images;

[0010] an image data processing module, the image data processing module being responsive to the image database and being configured to process the collected ground penetrating radar image, including performing image format conversion and / or size adjustment on the ground penetrating radar image;

[0011] Processing the collected ground penetrating radar image also includes dividing the collected ground penetrating radar image into ▽1, ▽2, ..., ▽ m , where ▽ m Representing the mth divided ground penetrating radar image, and obtaining an edge contour image corresponding to the ground penetrating radar image in each of the divided ground penetrating radar images;

[0012] a noise detection module, responsive to the edge contour image, configured to identify and detect a noise type and / or noise level corresponding to the acquired edge contour image, and acquire data related to the noise type, wherein the data includes distinguishing environmental noise, device noise, and other interfering noise according to the noise type; and the noise level represents the level of noise.

[0013] a noise reduction processing module, the noise reduction processing module being responsive to the acquired noise type and / or noise level and configured to select a noise reduction algorithm to perform noise reduction processing on the ground penetrating radar image;

[0014] An image fusion processing module is used to perform quality assessment on the ground penetrating radar image after noise reduction processing, and the image fusion processing module includes a comparison unit, a calculation unit and a determination unit;

[0015] The comparison unit is used to compare the sizes of the edge contour images, analyze the change in resolution of the edge contour images of different sizes after noise reduction processing, and obtain the original resolution of each edge contour image according to the change.

[0016] As a preferred solution of the present invention, the calculation unit is used to obtain the resolution of each edge contour image after noise reduction processing based on the original resolution of each edge contour image, and calculate the difference between the resolution and the original resolution according to the resolution. If the difference between the resolution and the original resolution is greater than 5%, it indicates that the difference is high, otherwise, the difference is low.

[0017] As a preferred solution of the present invention, the judgment module responds to the calculated difference, and obtains the corresponding edge contour image when the difference between the calculated resolution and the original resolution is high, and the system determines that the quality of the noise reduction processing of the corresponding edge contour image by the selected noise reduction algorithm is high; otherwise, no judgment is made.

[0018] As a preferred solution of the present invention, in which: in the noise reduction processing module, the selected noise reduction algorithms include wavelet transform noise reduction algorithm, filtering noise reduction algorithm and deep learning-based noise reduction algorithm. If the system determines that the selected noise reduction algorithm has high quality of noise reduction processing for the corresponding edge contour image, then the noise reduction algorithm is selected when performing noise reduction processing on the edge contour image, and the noise reduction algorithm with high noise reduction processing quality is used as the processing algorithm for the edge contour image.

[0019] As a preferred solution of the present invention, wherein: in the judgment module, when the difference between the resolution and the original resolution is low, the edge contour image is decomposed, and the decomposition method includes equally decomposing the edge contour image according to the area of ​​the edge contour image, obtaining the original resolution of the different decomposed parts, and performing noise reduction processing on each decomposed part based on the noise reduction algorithm, and calculating the change in resolution of each decomposed part after the noise reduction processing.

[0020] As a preferred solution of the present invention, the edge contour image is decomposed, and the decomposition method further includes counting the number of different resolutions appearing in the edge contour image, and different resolutions are divided into θ1, θ2, ..., θ n , where θ n Indicates the nth resolution distinguished, obtains the part below the average resolution in the distinguished resolution, performs noise reduction processing on the part by using a noise reduction algorithm, and if the resolution of the each part exceeds the average resolution, uses the noise reduction algorithm with high noise reduction processing quality as the processing algorithm of the edge contour image to reduce the noise of the edge contour image.

[0021] As a preferred solution of the present invention, if the resolution of the part does not exceed and / or is equal to the average resolution, the complex structural features of the edge contour image corresponding to this part are obtained, and the complex structural features include jagged and / or curved edge contours. The reference noise reduction algorithm is adjusted in the jagged and / or curved edge contours, including adjusting the filtering parameters, the filtering parameters including the standard deviations in the spatial domain and the color domain, and the standard deviation is increased until the resolution of the jagged and / or curved edge contour exceeds and / or is equal to the average resolution.

[0022] As a preferred solution of the present invention, the reference noise reduction algorithm is adjusted in the jagged and / or curved edge contour, and the adjustment method also includes hierarchical decomposition of the jagged and / or curved edge contour, including decomposition into a background layer, an edge layer and a detail layer, and noise reduction processing is performed on the edge contours of different levels until the resolution of the jagged and / or curved edge contour exceeds and / or is equal to the average resolution.

[0023] As a preferred solution of the present invention, if the resolution of the jagged and / or curved edge contour exceeds and / or is equal to the average resolution, the amplitude of the increase and / or decrease of the standard deviation is recorded, and the amplitude is marked as the reference amplitude. When the edge contour image is denoised in the future, if the edge contour image has a jagged and / or curved edge contour, the edge contour image is algorithmically adjusted with the same amplitude as the reference amplitude.

[0024] Beneficial effects:

[0025] 1. By analyzing edge contour images, the system can accurately distinguish between environmental noise, equipment noise, and other interfering noises, providing a basis for targeted noise reduction. During the noise reduction process, the system detects the quality of edge contour images and uses a variety of algorithms and strategies to effectively remove noise while preserving the image's edges and details to the greatest extent possible.

[0026] 2. The system can record and learn effective noise reduction algorithms and parameter settings, providing a reference for future noise reduction processing of similar images, thereby improving the system's adaptability and processing efficiency;

[0027] 3. The system can effectively process complex edge contour images, improve the system's adaptability to different scenarios and conditions, enable the system to dynamically adjust the processing strategy according to the complexity and noise level of the image, and enhance the flexibility and robustness of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. Those skilled in the art can also derive other drawings based on these drawings without inventive effort. Among them:

[0029] Figure 1 Schematic diagram of the modular structure of an intelligent noise reduction processing system for underground pipeline images based on ground penetrating radar according to an embodiment of the present invention;

[0030] Figure 2 This is a schematic diagram of the process structure of an embodiment of the present invention;

[0031] Figure 3 is a schematic diagram of an original image without noise reduction according to an embodiment of the present invention;

[0032] Figure 4 is a schematic diagram after noise reduction processing according to an embodiment of the present invention;

[0033] Reference numerals in the figure: 110 - ground penetrating radar image acquisition module; 120 - image data processing module; 130 - noise detection module; 140 - noise reduction processing module; 150 - image fusion processing module; 1501 - comparison unit; 1502 - calculation unit; 1503 - determination unit. DETAILED DESCRIPTION

[0034] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the described embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field are within the scope of protection of the present invention.

[0035] Due to the problems of modal aliasing and reliance on experience in existing technologies, it is difficult to accurately grasp the algorithm adjustments for different signal characteristics and / or shapes with complex structures. As a result, the image will be over-smoothed while removing noise, resulting in the loss of useful image details and affecting the accurate detection of the internal structure of underground pipelines.

[0036] Based on this, the present invention proposes an intelligent noise reduction processing system for underground pipeline images based on ground penetrating radar. It improves the noise reduction effect and image quality of ground penetrating radar images through noise detection and noise reduction processing, intelligent algorithm selection and optimization, complex structure processing and other means, providing more accurate and reliable image data support for the detection and maintenance of underground pipelines.

[0037] The present invention will be further described in detail below through embodiments and in conjunction with the accompanying drawings.

[0038] Reference Figures 1 to 4 , is an embodiment of the present invention, which provides an intelligent noise reduction processing system for underground pipeline images based on ground penetrating radar, comprising:

[0039] A ground penetrating radar image acquisition module 110 is used to acquire ground penetrating radar images of a preset underground pipeline infrastructure and to construct an image database based on the acquired ground penetrating radar images;

[0040] In this embodiment, an image database is constructed by collecting ground penetrating radar images, which provides a rich data foundation for subsequent image processing and analysis, and facilitates data management and mining;

[0041] An image data processing module 120 , which is responsive to the image database and is used to process the collected GPR images, including performing image format conversion and / or size adjustment on the GPR images;

[0042] Processing the collected ground penetrating radar image also includes dividing the collected ground penetrating radar image into ▽1, ▽2, ..., ▽ m , where ▽ m Representing the mth divided ground penetrating radar image, and obtaining the edge contour image of the ground penetrating radar image in each divided ground penetrating radar image;

[0043] Obtaining edge contour images of ground penetrating radar images during image processing provides feature information for noise detection and noise reduction, helping to more accurately identify noise types and perform targeted noise reduction, while also helping to protect important structural information of the image.

[0044] In this embodiment, edge contour images of ground-penetrating radar images are obtained. This edge information can reflect important features such as the interface between different media in the image and the boundary of the target object. In ground-penetrating radar images, edge contour information is crucial for understanding underground structures and target features.

[0045] At the same time, edge contour images can help distinguish noise from signals. If the edge contour becomes blurred or discontinuous under noise interference, it means that the noise may have a significant impact on the image details. For example, Gaussian noise will make the edge contour blurred, while impulse noise may cause artifacts or breaks in the edge contour.

[0046] Moreover, by analyzing the intensity changes of edge contours, the type and intensity of noise can be determined. For example, Gaussian noise will uniformly weaken the edge intensity, while impulse noise may cause local mutations in edge intensity.

[0047] The noise detection module 130 is responsive to the edge contour image and is configured to identify and detect the noise type and / or noise level corresponding to the acquired edge contour image, and obtain data related to the noise type, including distinguishing environmental noise, device noise, and other interfering noise according to the noise type; the noise level is the level of noise;

[0048] In this embodiment, the noise type and noise level corresponding to the edge contour image can be identified and detected, and relevant data can be obtained, which provides a basis for selecting an appropriate noise reduction algorithm and improves the pertinence and effectiveness of the noise reduction process.

[0049] The noise reduction processing module 140 is used to select a corresponding noise reduction algorithm to perform noise reduction processing on the ground penetrating radar image according to the noise type and / or noise level of the edge contour image;

[0050] The image fusion processing module 150 is used to perform quality assessment on the ground penetrating radar image after noise reduction processing. The image fusion processing module 150 includes a comparison unit 1501, a calculation unit 1502 and a determination unit 1503;

[0051] It should be noted that the image quality assessment in this embodiment is to perform a quality assessment on the GPR image after noise reduction, including evaluating indicators such as the signal-to-noise ratio (SNR) and peak signal-to-noise ratio (PSNR) of the GPR image;

[0052] Furthermore, quality assessment of the denoised GPR images can objectively evaluate the denoising effect and provide feedback for system optimization and improvement.

[0053] The comparison unit 1501 is used to compare the sizes of the edge contour images, analyze the changes in the resolution of the edge contour images of different sizes after the noise reduction process, and obtain the original resolution of each edge contour image according to the changes;

[0054] The calculation unit 1502 is used to obtain the resolution of each edge contour image after noise reduction based on the original resolution of each edge contour image, and calculate the difference between the resolution and the original resolution according to the resolution. If the difference between the resolution and the original resolution is greater than 5%, it indicates that the difference is high, otherwise, the difference is low.

[0055] In this embodiment, the calculation unit obtains the resolution of each edge contour image after noise reduction based on the original resolution of each edge contour image, and calculates the difference with the original resolution. This provides a quantitative indicator for evaluating the impact of noise reduction on image resolution, which helps to more accurately judge the effectiveness of the noise reduction algorithm.

[0056] The determination module 1503 responds to the calculated difference value, and when the difference between the calculated resolution and the original resolution is high, the corresponding edge contour image is obtained, and the system determines that the quality of the noise reduction processing of the corresponding edge contour image by the selected noise reduction algorithm is high; otherwise, no determination is made;

[0057] It should be noted that in this embodiment, the system determines that the quality of the noise reduction processing of the corresponding edge contour image by the selected noise reduction algorithm is high, which means that the difference between the resolution and the original resolution is greater than 10 centimeters;

[0058] The quality of the selected denoising algorithm for edge contour images is determined based on the calculated resolution difference. A high difference indicates a greater impact on image resolution, whereas a low difference indicates high denoising quality. This determination method provides a clear basis for system decision-making and helps select a more appropriate denoising algorithm.

[0059] In the noise reduction processing module, the selected noise reduction algorithms include a wavelet transform noise reduction algorithm, a filtering noise reduction algorithm, and a noise reduction algorithm based on deep learning. If the system determines that the selected noise reduction algorithm has high noise reduction processing quality for the corresponding edge contour image, then the noise reduction algorithm is selected when performing noise reduction processing on the edge contour image, and the noise reduction algorithm with high noise reduction processing quality is used as the processing algorithm for the edge contour image;

[0060] In this embodiment, edge contour images with good noise reduction effects are marked as reference edge contour images, providing a reference for future noise reduction processing of the same or similar edge contour images, thus achieving self-learning and optimization of the system and improving processing efficiency and quality.

[0061] On the basis of the above, the determination module further includes decomposing the edge contour image when the difference between the resolution and the original resolution is low, wherein the decomposition method includes equally decomposing the edge contour image according to the area of ​​the edge contour image, obtaining the original resolution of the different decomposed parts, performing noise reduction processing on each decomposed part based on a noise reduction algorithm, calculating the change in resolution of each decomposed part after the noise reduction processing, and obtaining an average resolution;

[0062] In this embodiment, the edge contour image is decomposed before the noise reduction process is performed. The core idea of ​​this method is to decompose a complex image into multiple simpler components, perform targeted noise reduction on each of these components, and then reassemble them.

[0063] In reality, different image components may be affected by different types of noise. By decomposing, each component can be analyzed and processed separately, and the most suitable noise reduction algorithm can be selected. For example, the high-frequency part may mainly contain details and noise, while the low-frequency part mainly contains background information and noise.

[0064] At the same time, directly denoising the entire edge contour image may lead to misjudgment, especially when the noise and signal features are similar. After decomposition, the features of noise and signal can be more clearly identified, thus reducing misjudgment;

[0065] Moreover, decomposition can better protect important features of the image (such as edges and details). For example, when processing high-frequency parts, a more aggressive noise reduction algorithm can be used, while when processing low-frequency parts, a more gentle algorithm can be used, thereby removing noise while retaining important information.

[0066] When the system does not determine that the selected noise reduction algorithm has high noise reduction processing quality for the edge contour image, the edge contour image is decomposed and noise reduction processing and resolution change calculation are performed on the decomposed parts separately. This decomposition processing method can analyze each part of the image more carefully, help to find problems in the noise reduction processing, and provide more specific information for further optimizing the noise reduction algorithm;

[0067] By calculating the resolution change of the decomposed parts, and calculating the total resolution and average resolution based on the total number, as well as counting the number and difference of parts with low original resolution and improved resolution after noise reduction, the effect of noise reduction processing is analyzed in a more detailed and quantitative manner, providing a more accurate basis for system optimization.

[0068] Furthermore, the edge contour image is decomposed, and the decomposition method also includes counting the number of different resolutions appearing in the edge contour image, and dividing the different resolutions into θ1, θ2, ..., θ n , where θ n Indicates the nth resolution that has been distinguished, obtains a portion of the distinguished resolution that is lower than the average resolution, and performs noise reduction processing on the portion using a noise reduction algorithm. If the resolution of the portion exceeds the average resolution, the noise reduction algorithm with the highest noise reduction quality is used as the processing algorithm for the edge contour image to perform noise reduction on the edge contour image;

[0069] In this embodiment, the decomposition method of the edge contour image is further enriched. In addition to decomposing by equal area, it also distinguishes and processes according to the number of statistically analyzed images with different resolutions. This resolution-based decomposition method can more accurately perform noise reduction processing on parts with different resolutions, thereby improving the pertinence and effectiveness of the noise reduction processing.

[0070] For the parts below the average resolution, the noise reduction algorithm is used for noise reduction processing. The selection of the noise reduction algorithm for edge contour images with the same resolution in the future is determined based on whether the processed resolution reaches the average resolution. This realizes the adaptive selection of the noise reduction algorithm and improves the intelligence level of the system.

[0071] Specifically, in this embodiment, if the resolution of a portion does not exceed the average resolution, complex structural features of the edge contour image corresponding to the portion are obtained, the complex structural features including jagged and / or curved edge contours, and a noise reduction algorithm is adjusted for the jagged and / or curved edge contours, including adjusting filtering parameters, including standard deviations in the spatial domain and the color domain, and increasing and / or decreasing the standard deviation until the resolution of the jagged and / or curved edge contour exceeds and / or equals the average resolution.

[0072] In this embodiment, regions with complex structures typically contain more details and variations, such as edges, textures, and noise. These details may be difficult to clearly capture and distinguish at a limited resolution, resulting in lower visual resolution. Complex structures are also more susceptible to interference from noise, leading to reduced image quality. Noise can mask or distort structural details, making the resolution appear lower. Therefore, this embodiment acquires structural features and makes corresponding algorithm adjustments, which has practical significance.

[0073] In this embodiment, the noise reduction algorithm is adjusted to target complex structural features such as jagged and / or curved edge contours, including adjusting the filter parameters. This special processing method for complex structures can better meet the noise reduction requirements of complex images, improve the quality of the noise reduction process, and improve the protection of complex structures.

[0074] By increasing and / or decreasing the standard deviation of the spatial and color domains in the filtering parameters until the resolution of the complex structure reaches the average resolution, a flexible adjustment method is provided for the optimization of the noise reduction algorithm, thereby enhancing the adaptability and robustness of the system.

[0075] Further, the noise reduction algorithm is adjusted in the jagged and / or curved edge contour, wherein the adjustment method further includes decomposing the jagged and / or curved edge contour into a background layer, an edge layer, and a detail layer, and performing noise reduction processing on the edge contours at different layers until the resolution of the jagged and / or curved edge contour exceeds and / or equals the average resolution;

[0076] In this embodiment, the jagged and / or curved edge contours are decomposed into a hierarchical layer, including a background layer, an edge layer, and a detail layer, and noise reduction processing is performed on each layer separately. This hierarchical decomposition method can process each component of the image in more detail, helping to better protect important image structures and detail information, and further improving the quality and effectiveness of the noise reduction processing;

[0077] On the basis of the above, if the resolution of the jagged and / or curved edge contour exceeds and / or is equal to the average resolution, then the amplitude of the increase and / or decrease of the standard deviation is recorded and marked as the reference amplitude. When performing noise reduction on the edge contour image in the future, if the edge contour image has a jagged and / or curved edge contour, the edge contour image is algorithmically adjusted by the same amplitude as the reference amplitude.

[0078] In this embodiment, the recording and reuse of parameter adjustments are implemented, which provides support for the self-learning and optimization of the system and improves the processing efficiency and quality stability.

[0079] In summary, the present invention constructs a complete intelligent noise reduction processing system for underground pipeline images based on GPR, covering the entire process from image acquisition, data processing, noise detection to noise reduction processing and quality assessment, realizing systematic noise reduction processing of GPR images, and providing a comprehensive solution for underground pipeline detection.

[0080] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. An intelligent noise reduction processing system for underground pipeline images based on ground penetrating radar, characterized by: include: A ground penetrating radar image acquisition module, used to acquire ground penetrating radar images of the infrastructure of the preset underground pipeline and to construct an image database based on the acquired ground penetrating radar images; an image data processing module, the image data processing module being responsive to the image database and being configured to process the collected ground penetrating radar image, including performing image format conversion and / or size adjustment on the ground penetrating radar image; Processing the collected ground penetrating radar image also includes dividing the collected ground penetrating radar image into ▽1, ▽2, ..., ▽ m , where ▽ m Representing the mth divided ground penetrating radar image, and obtaining an edge contour image of the ground penetrating radar image in each of the divided ground penetrating radar images; a noise detection module, responsive to the edge contour image, configured to identify and detect a noise type and / or noise level corresponding to the acquired edge contour image, and acquire data related to the noise type, wherein the data includes distinguishing environmental noise, device noise, and other interfering noise according to the noise type; and the noise level represents the level of noise. A noise reduction processing module is used to select a corresponding noise reduction algorithm to perform noise reduction processing on the ground penetrating radar image according to the noise type and / or noise level of the edge contour image; An image fusion processing module is used to perform quality assessment on the ground penetrating radar image after noise reduction processing, and the image fusion processing module includes a comparison unit, a calculation unit and a determination unit; The comparison unit is used to compare the sizes of the edge contour images, analyze the change in resolution of the edge contour images of different sizes after noise reduction processing, and obtain the original resolution of each edge contour image according to the change.

2. The intelligent noise reduction processing system for underground pipeline images based on ground penetrating radar according to claim 1, characterized in that: The calculation unit is used to obtain the resolution of each edge contour image after noise reduction processing based on the original resolution of each edge contour image, and calculate the difference between the resolution and the original resolution according to the resolution. If the difference between the resolution and the original resolution is greater than 5%, it indicates that the difference is high, otherwise, the difference is low.

3. The intelligent noise reduction processing system for underground pipeline images based on ground penetrating radar according to claim 1, characterized in that: The determination module responds to the calculated difference, and obtains the corresponding edge contour image when the difference between the calculated resolution and the original resolution is high, and the system determines that the quality of the noise reduction processing of the corresponding edge contour image by the selected noise reduction algorithm is high; otherwise, no determination is made.

4. The intelligent noise reduction processing system for underground pipeline images based on ground penetrating radar according to claim 3, characterized in that: In the noise reduction processing module, the selected noise reduction algorithms include wavelet transform noise reduction algorithm, filtering noise reduction algorithm and deep learning-based noise reduction algorithm. If the system determines that the selected noise reduction algorithm has high quality of noise reduction processing for the corresponding edge contour image, then the noise reduction algorithm is selected when performing noise reduction processing on the edge contour image, and the noise reduction algorithm with high noise reduction processing quality is used as the processing algorithm for the edge contour image.

5. The intelligent noise reduction processing system for underground pipeline images based on ground penetrating radar according to claim 4, characterized in that: In the judgment module, when the difference between the resolution and the original resolution is low, the edge contour image is decomposed. The decomposition method includes equally decomposing the edge contour image according to the area of ​​the edge contour image, obtaining the original resolution of the different decomposed parts, and performing noise reduction processing on each decomposed part based on the noise reduction algorithm. After the noise reduction processing, the change in the resolution of each decomposed part is calculated, and the average resolution is obtained.

6. The intelligent noise reduction processing system for underground pipeline images based on ground penetrating radar according to claim 5, characterized in that: The edge contour image is decomposed, and the decomposition method further includes counting the number of different resolutions appearing in the edge contour image, and different resolutions are divided into θ1, θ2, ..., θ n , where θ n Indicates the nth resolution distinguished, obtains the part below the average resolution in the distinguished resolution, performs noise reduction processing on the part by referring to the noise reduction algorithm, and if the resolution of each part exceeds the average resolution, uses the noise reduction algorithm with high noise reduction processing quality as the processing algorithm of the edge contour image to reduce the noise of the edge contour image.

7. The intelligent noise reduction processing system for underground pipeline images based on ground penetrating radar according to claim 6, characterized in that: If the resolution of the part does not exceed and / or is equal to the average resolution, the complex structural features of the edge contour image corresponding to this part are obtained, and the complex structural features include jagged and / or curved edge contours. The noise reduction algorithm is adjusted in the jagged and / or curved edge contours, including adjusting the filtering parameters, the filtering parameters including the standard deviations in the spatial domain and the color domain, and the standard deviation is increased until the resolution of the jagged and / or curved edge contour exceeds and / or is equal to the average resolution.

8. The intelligent noise reduction processing system for underground pipeline images based on ground penetrating radar according to claim 7, characterized in that: The noise reduction algorithm is adjusted in the jagged and / or curved edge contour, and the adjustment method also includes hierarchical decomposition of the jagged and / or curved edge contour, including decomposition into a background layer, an edge layer and a detail layer, and noise reduction processing is performed on the edge contours of different levels until the resolution of the jagged and / or curved edge contour exceeds and / or is equal to the average resolution.

9. The intelligent noise reduction processing system for underground pipeline images based on ground penetrating radar according to claim 7, characterized in that: If the resolution of the jagged and / or curved edge contour exceeds and / or is equal to the average resolution, the amplitude of the increase and / or decrease in the standard deviation is recorded and the amplitude is marked as the reference amplitude. When denoising the edge contour image in the future, if the edge contour image has a jagged and / or curved edge contour, the edge contour image is algorithmically adjusted with the same amplitude as the reference amplitude.

Images