Roadbed data denoising method and device

Abstract

The application discloses a kind of roadbed data denoising method and device, wherein the method comprises: obtaining the roadbed data with noise detected by radar;The roadbed data with noise is blocked, and reference block is extracted;According to the preset threshold, the corresponding matching block of reference block is determined, the matching block is added to the matching block array, the corresponding matching block array and the estimated value of reference block are obtained;After the corresponding matching block array of reference block is filtered jointly, all estimated values of reference block are returned to the corresponding preset position;All estimated values of reference block returned to the corresponding preset position are weighted and averaged, and the value after roadbed data denoising is obtained.The application can reduce noise interference and improve signal channel signal-to-noise ratio, effectively distinguish high-frequency interference caused by noise.

Description

Methods and devices for denoising roadbed data Technical Field

[0001] This invention relates to the field of signal processing technology, and in particular to a method and apparatus for denoising roadbed data. Background Technology

[0002] This section is intended to provide background or context for the embodiments of the invention set forth in the claims. The description herein is not an admission that it is prior art simply because it is included in this section.

[0003] Timely detection of railway subgrade defects, remediation of subgrade safety hazards, and maintenance of railway subgrade stability are prerequisites for safe railway operation. Currently, subgrade defects are widespread on existing railway lines, with subgrade subsidence and mud pumping being particularly serious. Traditional subgrade detection and noise reduction methods primarily rely on transform domain methods, using filters to remove external noise. However, these methods cannot effectively distinguish between high-frequency signals and high-frequency interference caused by noise, and also incur a very high computational burden. Summary of the Invention

[0004] This invention provides a method for denoising roadbed data to reduce noise interference and improve the signal-to-noise ratio of signal channels, thereby effectively distinguishing high-frequency interference caused by noise. The method includes:

[0005] Acquire noisy roadbed data detected by radar;

[0006] The noisy roadbed data is divided into blocks, and reference blocks are extracted.

[0007] Based on a preset threshold, the matching block corresponding to the reference block is determined, and the matching block is added to the matching block array to obtain the corresponding matching block array and estimated value of the reference block.

[0008] After performing joint filtering on the corresponding matching block array of the reference block, all estimated values ​​of the reference block are returned to the corresponding preset positions;

[0009] A weighted average is performed on all estimated values ​​of the reference block returned to the corresponding preset position to obtain the denoised values ​​of the roadbed data.

[0010] This invention also provides a roadbed data denoising device to reduce noise interference and improve the signal-to-noise ratio of signal channels, thereby effectively distinguishing high-frequency interference caused by noise. The device includes:

[0011] The data acquisition module is used to acquire noisy roadbed data detected by radar;

[0012] The block operation module is used to perform block operations on noisy roadbed data and extract reference blocks;

[0013] The matching block array acquisition module is used to determine the matching block corresponding to the reference block according to a preset threshold, add the matching block to the matching block array, and obtain the corresponding matching block array and estimated value of the reference block;

[0014] The preset position return module is used to perform joint filtering on the corresponding matching block array of the reference block and return all the estimated values ​​of the reference block to the corresponding preset position.

[0015] The data denoising module is used to perform weighted averaging on all estimated values ​​of the reference block returned to the corresponding preset position to obtain the denoised values ​​of the roadbed data.

[0016] This invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the above-described roadbed data denoising method.

[0017] This invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described roadbed data denoising method.

[0018] This invention also provides a computer program product, which includes a computer program that, when executed by a processor, implements the above-described roadbed data denoising method.

[0019] In this embodiment of the invention, noisy roadbed data detected by radar is acquired; the noisy roadbed data is segmented to extract reference blocks; matching blocks corresponding to the reference blocks are determined according to a preset threshold, and the matching blocks are added to a matching block array to obtain the corresponding matching block array and estimated values ​​of the reference blocks; after joint filtering of the corresponding matching block array of the reference blocks, all estimated values ​​of the reference blocks are returned to the corresponding preset positions; a weighted average is performed on all estimated values ​​of the reference blocks returned to the corresponding preset positions to obtain the denoised values ​​of the roadbed data. In the above process, this embodiment of the invention segments the noisy roadbed data to obtain reference blocks, corresponding matching block arrays, and estimated values, and performs a weighted average on all estimated values ​​of the reference blocks returned to the corresponding preset positions, thereby reducing noise interference and improving the signal-to-noise ratio of the signal channel, effectively distinguishing high-frequency interference caused by noise. Attached Figure Description

[0020] 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. In the drawings:

[0021] Figure 1 is a flowchart of the roadbed data denoising method in an embodiment of the present invention;

[0022] Figure 2 is a flowchart of the weighted average processing of the estimated values ​​in an embodiment of the present invention;

[0023] Figure 3 shows the roadbed data containing noise in an embodiment of the present invention;

[0024] Figure 4 shows the residual data in an embodiment of the present invention;

[0025] Figure 5 shows the denoised roadbed data in an embodiment of the present invention;

[0026] Figure 6 is a schematic diagram of the roadbed data denoising device in an embodiment of the present invention. Detailed Implementation

[0027] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings. Here, the illustrative embodiments of the present invention and their descriptions are used to explain the present invention, but are not intended to limit the present invention.

[0028] Figure 1 is a flowchart of the roadbed data denoising method in an embodiment of the present invention, including:

[0029] Step 101: Obtain noisy roadbed data detected by radar;

[0030] Step 102: Perform block segmentation on the noisy roadbed data and extract reference blocks;

[0031] Step 103: Determine the matching block corresponding to the reference block according to the preset threshold, add the matching block to the matching block array, and obtain the corresponding matching block array and estimated value of the reference block;

[0032] Step 104: After performing joint filtering on the corresponding matching block array of the reference block, return all estimated values ​​of the reference block to the corresponding preset position;

[0033] Step 105: Perform a weighted average of all estimated values ​​of the reference block returned to the corresponding preset position to obtain the denoised values ​​of the roadbed data.

[0034] The following explains each step in detail.

[0035] In step 101, noisy roadbed data detected by radar is acquired.

[0036] In specific embodiments, ground-penetrating radar, with its advantages of high-speed non-destructive testing, convenient operation, and high efficiency, has become a commonly used tool for roadbed data detection.

[0037] In step 102, the noisy roadbed data is divided into blocks to extract reference blocks.

[0038] In one embodiment, after extracting the reference block, the method further includes:

[0039] The roadbed data in the reference block is normalized.

[0040] In a specific embodiment, the roadbed data is divided into blocks to obtain two-dimensional reference blocks, and then a three-dimensional array is generated by grouping and arranging them. The noisy roadbed data consists of the following form:

[0041] D(x)=S(x)+δ(x), x∈X

[0042] Where x is the profile data. The spatial coordinates of δ, and δ is in the form of σ. 2 It is zero-mean Gaussian noise with zero variance, i.e., δ(·)~N(0,σ) 2 ).

[0043] We take a block of fixed size N1×N1 to represent D(x). x Where x is the coordinate of the block. A two-dimensional reference block group can be represented by a bold uppercase letter with a subscript that is the set of coordinates of the blocks in its group, for example, D S It is a company located in Block D x The array is constructed, and the data of size D is extracted sequentially from the noisy roadbed data D. (For example Reference block (located at current coordinate x) R ∈D), using a sliding window approach, reference blocks are randomly extracted within the range of noisy roadbed data. For reference block The roadbed data was normalized using the normalization algorithm Ψ(·), where... in

[0044] In step 103, a matching block corresponding to the reference block is determined according to a preset threshold, and the matching block is added to the matching block array to obtain the corresponding matching block array and estimated value of the reference block.

[0045] In one embodiment, determining the matching block corresponding to the reference block based on a preset threshold includes:

[0046] When the reference distance between the reference block and the matching block is less than a preset threshold, the matching block is determined to be the matching block of the reference block.

[0047] In one embodiment, adding the matching block to the matching block array includes:

[0048] When the number of matching blocks in the matching block array does not exceed the current upper limit of matching blocks, the matching block is added to the matching block array.

[0049] In a specific embodiment, the lookup and reference block corresponding matching block They are then stacked together to form a matching block array. A preset threshold ε is set; if the reference block... Matching blocks The distance between dis(x) R ,x T If the value is less than the preset threshold ε, then the matching block is... The matching block will be identified as the reference block, and the matching block will be grouped into the same group. Reference block Matching blocks The distance between dis(x) R ,x T Using L2 distance as the metric, that is:

[0050]

[0051] in, express The number of elements in the middle.

[0052] For any If the distance is dis(x) R ,x T If )≤ε, then the matching block The corresponding matching block array added to the reference block forms the reference block. Corresponding matching block array make Represents a matching array The number of elements in the group, setting the upper limit for the number of blocks in the group. Right now To limit the size of the group.

[0053] In step 104, after performing joint filtering on the corresponding matching block array of the reference block, all estimated values ​​of the reference block are returned to the corresponding preset positions.

[0054] In one embodiment, the combined filtering process includes one or any combination of: matched block array 3D transformation, transform spectrum contraction, and inverse transformation.

[0055] Joint filtering is performed on the matched block array, and all estimated values ​​of the reference block are returned to the preset position. Due to the obtained reference block... The estimates may overlap, so each data point contains multiple estimates. Matching block array Stack them to form a size of The three-dimensional array data, denoted as Where the matching block array The order of the elements can be randomly arranged to form a three-dimensional array.

[0056] For all The array employs a three-dimensional domain transform, attenuating noise by hard-thresholding the transform coefficients, and then inverts the three-dimensional domain transform to generate all reference blocks. The estimated value forms the estimation block and block It is estimated that it will return to its preset position.

[0057] In a specific embodiment, Array performing 3D discrete cosine transform This is performed in the transform domain by pre-setting a hard threshold δ. 3D =0.05 filtering, that is, if make Then inverse transformation In the spatial domain, obtain Estimated block after collaborative hard thresholding

[0058] In step 105, a weighted average is performed on all estimated values ​​of the reference block returned to the corresponding preset position to obtain the denoised values ​​of the roadbed data.

[0059] Figure 2 is a flowchart of the weighted average processing of the estimated values ​​in an embodiment of the present invention. In one embodiment, the weighted average processing of all estimated values ​​of the reference block returned to the corresponding preset position includes:

[0060] Step 201: Assign a weight to all estimated values ​​for the same preset location;

[0061] Step 202: Based on the weights, perform a weighted average calculation on all estimated values ​​of the reference block returning to the corresponding preset position to obtain a weighted average result of all weights.

[0062] A preliminary estimate of the noisy data is calculated by weighting all the obtained overlapping block estimates.

[0063] In a specific embodiment, after joint filtering, the estimated block is placed back into its preset position, but due to the search for matching blocks... When there is block overlap, the preset position point x of the overlapping area is determined. R There will be more than one denoised result, requiring the same preset location point x. R Different estimates are assigned a weight, and the weighted average is taken as the final result. That is, for any point (x, y), the corresponding weighted average value is:

[0064]

[0065] Then, the filtered estimation blocks are put back into their preset positions to obtain the initial denoising results.

[0066] In a specific embodiment, the location of the matching block corresponding to the reference block is found within the range of noisy roadbed data. Based on this location information, an image of the noisy roadbed data is formed, as shown in Figure 3. The difference between the data in the reference block and the corresponding estimated value is the residual data, and Figure 4 shows the residual data in an embodiment of the present invention.

[0067] Based on the initial denoising result D * After processing, build Calculate the weighting coefficients:

[0068]

[0069] Then through the formula in, The corresponding convolution operation, calculate The estimated value Finally, the estimation results of the noisy data were obtained. The denoised roadbed data is shown in Figure 5.

[0070] In summary, the method proposed in the embodiments of the present invention has the following beneficial effects:

[0071] It reduces noise interference and improves the signal-to-noise ratio of the signal channel, effectively distinguishes high-frequency interference caused by noise, and preserves the basic unique characteristics of each data segment through the fusion technology of the spatiotemporal domain and the frequency domain.

[0072] This invention also provides a roadbed data denoising device, as described in the following embodiments. Since the principle of this device in solving the problem is similar to that of the roadbed data denoising method, the implementation of this device can refer to the implementation of the roadbed data denoising method, and repeated details will not be elaborated further. Figure 6 is a schematic diagram of the roadbed data denoising device in this invention, including:

[0073] The data acquisition module 601 is used to acquire noisy roadbed data detected by radar;

[0074] The block operation module 602 is used to perform block operations on noisy roadbed data and extract reference blocks;

[0075] The matching block array acquisition module 603 is used to determine the matching block corresponding to the reference block according to a preset threshold, add the matching block to the matching block array, and obtain the corresponding matching block array and estimated value of the reference block;

[0076] The preset position return module 604 is used to return all estimated values ​​of the reference block to the corresponding preset position after performing joint filtering on the corresponding matching block array of the reference block;

[0077] The data denoising module 605 is used to perform weighted average processing on all estimated values ​​of the reference block returned to the corresponding preset position to obtain the denoised values ​​of the roadbed data.

[0078] In one embodiment, the block operation module 602 is further configured to:

[0079] The roadbed data in the reference block is normalized.

[0080] In one embodiment, the matching block array acquisition module 603 is specifically used for:

[0081] When the reference distance between the reference block and the matching block is less than a preset threshold, the matching block is determined to be the matching block of the reference block.

[0082] In one embodiment, the matching block array acquisition module 603 is further configured to:

[0083] When the number of matching blocks in the matching block array does not exceed the current upper limit of matching blocks, the matching block is added to the matching block array.

[0084] In one embodiment, the joint filtering process includes one or any combination of: matched block array 3D transformation, transform spectrum shrinkage, and inverse transformation.

[0085] In one embodiment, the data denoising module 605 is used for:

[0086] Assign a weight to all estimated values ​​for the same preset location;

[0087] Based on the weights, a weighted average is calculated on all estimated values ​​of the reference block returning to the corresponding preset position to obtain a weighted average result of all weights.

[0088] In summary, the device proposed in this embodiment of the invention has the following beneficial effects: reducing noise interference and improving the signal-to-noise ratio of the signal channel, effectively distinguishing high-frequency interference caused by noise, and preserving the basic unique characteristics of each data segment through the fusion technology of the spatiotemporal domain and the frequency domain.

[0089] This invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the above-described roadbed data denoising method.

[0090] This invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described roadbed data denoising method.

[0091] This invention also provides a computer program product, which includes a computer program that, when executed by a processor, implements the above-described roadbed data denoising method.

[0092] In this embodiment of the invention, noisy roadbed data detected by radar is acquired; the noisy roadbed data is segmented to extract reference blocks; matching blocks corresponding to the reference blocks are determined according to a preset threshold, and the matching blocks are added to a matching block array to obtain the corresponding matching block array and estimated values ​​of the reference blocks; after joint filtering of the corresponding matching block array of the reference blocks, all estimated values ​​of the reference blocks are returned to the corresponding preset positions; a weighted average is performed on all estimated values ​​of the reference blocks returned to the corresponding preset positions to obtain the denoised values ​​of the roadbed data. In the above process, this embodiment of the invention segments the noisy roadbed data to obtain reference blocks, corresponding matching block arrays, and estimated values, and performs a weighted average on all estimated values ​​of the reference blocks returned to the corresponding preset positions, thereby reducing noise interference and improving the signal-to-noise ratio of the signal channel, effectively distinguishing high-frequency interference caused by noise.

[0093] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0094] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in one or more blocks of the flowchart illustrations and / or one or more blocks of the block diagrams.

[0095] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means that implement the functions specified in one or more flowcharts and / or one or more block diagrams.

[0096] These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, such that the instructions, which execute on the computer or other programmable apparatus, provide steps for implementing the functions specified in one or more flowcharts and / or one or more block diagrams.

[0097] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for denoising roadbed data, characterized in that, include: Acquire noisy roadbed data detected by radar; The noisy roadbed data is divided into blocks, and reference blocks are extracted. Based on a preset threshold, a matching block corresponding to the reference block is determined, and the matching block is added to the matching block array to obtain the corresponding matching block array and estimated value of the reference block; after joint filtering of the corresponding matching block array of the reference block, all estimated values ​​of the reference block are returned to the corresponding preset position; a weighted average is performed on all estimated values ​​of the reference block returned to the corresponding preset position to obtain the denoised value of the roadbed data; Based on a preset threshold, the matching block corresponding to the reference block is determined, including: when the reference distance between the reference block and the matching block is less than the preset threshold, the matching block is determined as the matching block of the reference block; the matching block is added to the matching block array, including: when the number of matching blocks in the matching block array does not exceed the current upper limit of matching blocks, the matching block is added to the matching block array; after performing joint filtering on the corresponding matching block array of the reference block, all estimated values ​​of the reference block are returned to the corresponding preset positions, including: stacking the matching block arrays to form three-dimensional array data, wherein the order of the elements in the matching block array is randomly arranged to form a three-dimensional array; a three-dimensional domain transformation is applied to all three-dimensional arrays, and noise is attenuated by hard thresholding the transformation coefficients; the three-dimensional domain transformation is reversed to generate an estimated block formed by the estimated values ​​of all reference blocks, and the estimated block is returned to its preset position; a 3D discrete cosine transform is performed on the three-dimensional array in the transform domain, and filtering is performed by a preset hard threshold. If the transformed value is less than the hard threshold, the transformed value is set to 0, and then the inverse transform is performed to the spatial domain to obtain the estimated block of the reference block after collaborative hard threshold filtering.

2. The method as described in claim 1, characterized in that, After extracting the reference block, the process also includes normalizing the roadbed data in the reference block.

3. The method as described in claim 1, characterized in that, The joint filtering process includes one or any combination of: matched block array 3D transformation, transformation spectrum shrinkage, and inverse transformation.

4. The method as described in claim 1, characterized in that, The weighted average of all estimated values ​​of the reference block returning to the corresponding preset position is performed, including: assigning a weight to all estimated values ​​of the same preset position; and calculating a weighted average of all estimated values ​​of the reference block returning to the corresponding preset position based on the weight to obtain a weighted average result of all weights.

5. A roadbed data denoising device, characterized in that, include: The data acquisition module is used to acquire noisy roadbed data detected by radar; The block segmentation module is used to segment the noisy roadbed data into blocks and extract reference blocks; the matching block array acquisition module is used to determine the matching block corresponding to the reference block according to a preset threshold, add the matching block to the matching block array, and obtain the corresponding matching block array and estimated value of the reference block. The preset position return module is used to perform joint filtering on the corresponding matching block array of the reference block and return all the estimated values ​​of the reference block to the corresponding preset position. The data denoising module is used to perform weighted average processing on all estimated values ​​of the reference block returned to the corresponding preset position to obtain the denoised value of the roadbed data; the matching block array acquisition module is specifically used to: determine the matching block as the matching block of the reference block when the reference distance between the reference block and the matching block is less than a preset threshold; the matching block array acquisition module is also used to: add the matching block to the matching block array when the number of matching blocks in the matching block array does not exceed the current upper limit of matching blocks; The preset position return module is also used to: stack the matching block array to form three-dimensional array data, wherein the order of the elements in the matching block array is randomly arranged to form a three-dimensional array; A three-dimensional domain transformation is applied to all three-dimensional arrays. Noise is attenuated by hard thresholding the transformation coefficients. The three-dimensional domain transformation is reversed to generate an estimation block formed by the estimated values ​​of all reference blocks. The estimation block is then returned to its preset position. The 3D array is subjected to a 3D discrete cosine transform in the transform domain. Filtering is performed by a pre-set hard threshold. If the transformed value is less than the hard threshold, the transformed value is set to 0. Then, the inverse transform is performed to the spatial domain to obtain the estimated block of the reference block after collaborative hard thresholding.

6. The apparatus as claimed in claim 5, characterized in that, The block operation module is also used to normalize the roadbed data in the reference block.

7. The apparatus as claimed in claim 5, characterized in that, The joint filtering process includes one or any combination of: matched block array 3D transformation, transformation spectrum shrinkage, and inverse transformation.

8. The apparatus as claimed in claim 5, characterized in that, The data denoising module is specifically used to: assign a weight to all estimated values ​​at the same preset position; and, based on the weight, perform a weighted average calculation on all estimated values ​​of the reference block that return to the corresponding preset position to obtain a weighted average result of all weights.

9. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method of any one of claims 1 to 5.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method of any one of claims 1 to 5.

11. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the method of any one of claims 1 to 5.

Images