A method and apparatus for retrieving and locating ground-penetrating radar images based on spatial prior constraints

Abstract

This application relates to a ground-penetrating radar (GPR) image retrieval and localization method and apparatus based on spatial prior constraints. This method improves localization accuracy by generally utilizing single-channel GPR data for similarity measurement and combining it with spatial prior constraints of the array. Specifically, firstly, multi-level wavelet transform is used to extract features based on the characteristics of GPR A-scan data, and a single-channel GPR data feature library is constructed. Then, because GPR data has limited information, the probability of mismatches is still relatively high when using only one-dimensional data for registration. Therefore, a spatial prior constraint matrix for the array channels is added to utilize fixed distance information between channels to determine valid retrieval pairs. This reduces the impact of mismatches caused by relying solely on similarity retrieval, effectively improving the localization accuracy of GPR in cases of incomplete maps, and providing important assurance for achieving safe and reliable GPR navigation and localization.

Description

Technical Field

[0001] This invention belongs to the field of ground-penetrating radar navigation and positioning technology, and relates to a ground-penetrating radar image retrieval and positioning method and device based on spatial prior constraints. Background Technology

[0002] Ground penetrating radar (GPR) transmits high-frequency electromagnetic waves into the ground and analyzes the reflected signals to obtain underground information. GPR is now widely used in underground target detection and road maintenance. With the continuous development of autonomous driving technology, GPR positioning systems offer a new research direction for existing navigation and positioning methods. Localizing Ground Penetrating Radar (LGPR) is a new positioning method that utilizes relatively stable underground features and achieves autonomous positioning through registration of underground data. Compared to optical and lidar-based driver assistance systems, LGPR positioning is less dependent on surface information and has the advantage of being less affected by weather and lighting conditions. As a new driver assistance positioning system, GPR provides a supplement to the field of autonomous driving, improving the robustness and accuracy of autonomous driving navigation and positioning. Furthermore, with the development of smart cities and the vigorous promotion of new infrastructure, LGPR systems have broad application prospects in industries such as urban rail transit safety, tunnels, coal mines, logistics, and manufacturing.

[0003] Current ground-penetrating radar (GPR) navigation and positioning technologies primarily focus on map-based matching. The aim is to register real-time acquired GPR images with pre-acquired map data to determine the system's current location. However, in developing this invention, the inventors discovered that traditional GPR positioning technology has high map requirements. While it performs well with complete maps, incomplete map data leads to limited GPR image information. This results in numerous mismatches when using traditional techniques for registration in low-overlap image scenarios, severely impacting navigation and positioning. In other words, there is a technical problem of insufficient GPR image positioning accuracy in low-overlap image situations. Summary of the Invention

[0004] To address the problems existing in the above-mentioned traditional methods, this invention proposes a ground-penetrating radar image retrieval and positioning method based on spatial prior constraints and a ground-penetrating radar image retrieval and positioning device based on spatial prior constraints, which can effectively improve the positioning accuracy of ground-penetrating radar images in the case of low-overlapping images.

[0005] To achieve the above objectives, the embodiments of the present invention adopt the following technical solutions:

[0006] On the one hand, a ground-penetrating radar image retrieval and localization method based on spatial prior constraints is provided, including the following steps:

[0007] Acquire a pre-built single-channel data feature library for multi-channel ground-penetrating radar systems;

[0008] A multi-channel ground-penetrating radar system was used to collect real-time two-dimensional ground-penetrating radar data, and multi-level wavelet transform was used to extract channel dimension data features from the two-dimensional ground-penetrating radar data.

[0009] Based on the characteristics of the channel dimension data, a similarity search is performed in the single-channel data feature library to obtain the channel position of each channel of the multi-channel ground penetrating radar system in the single-channel data feature library;

[0010] Calculate the retrieval channel distance matrix based on the location of each channel, and extract effective retrieval pairs using the constructed spatial prior constraint matrix and the retrieval channel distance matrix to obtain effective retrieval pairs.

[0011] After determining the list of candidate valid search channels based on the valid search pairs, the valid search channels are determined from the list of candidate valid search channels based on the Euclidean distance between the single channel data features in the channel dimension data features and the single channel data features in the single channel data feature library.

[0012] Based on the effective retrieval channels, a straight line fit is performed to calculate the position of the array center point of the multi-channel ground penetrating radar system, and the ground penetrating radar image retrieval and positioning results are obtained.

[0013] In one embodiment, the above-described ground-penetrating radar image retrieval and localization method based on spatial prior constraints further includes the following steps:

[0014] A multi-channel ground-penetrating radar system was used to acquire three-dimensional map data from ground-penetrating radar, resulting in three-dimensional volume data composed of multiple single-channel dimensional data.

[0015] Feature extraction is performed on each single-channel dimension data in the three-dimensional volume data to construct a single-channel data feature library for a multi-channel ground penetrating radar system.

[0016] In one embodiment, the step of extracting features from each single-channel dimension of the three-dimensional volume data to construct a single-channel data feature library for a multi-channel ground-penetrating radar system includes:

[0017] Multi-level wavelet decomposition is performed on each single-channel dimension data in the three-dimensional volume data in the frequency domain to obtain the single-channel data features of each single-channel dimension data, and a single-channel data feature library is obtained.

[0018] In one embodiment, the distance constraint between valid retrieval pairs during the process of obtaining valid retrieval pairs is:

[0019]

[0020] Where Λ represents the distance error threshold, and R ij The distance from the i-th channel to the j-th channel of the array antenna is represented by D, denoted as the spatial prior constraint matrix. ij Let represent the distance between the i-th channel and the j-th channel obtained through location retrieval, denoted as the retrieval channel distance matrix, and abs() represents calculating the absolute value.

[0021] On the other hand, a ground-penetrating radar image retrieval and positioning device based on spatial prior constraints is also provided, comprising:

[0022] The feature library acquisition module is used to acquire a pre-built single-channel data feature library for multi-channel ground penetrating radar systems;

[0023] The real-time extraction module is used to collect real-time 2D ground-penetrating radar data using a multi-channel ground-penetrating radar system, and to extract channel dimension data features from the 2D ground-penetrating radar data using multi-level wavelet transform.

[0024] The channel location module is used to perform a similarity search in the single-channel data feature library based on the channel dimension data features, so as to obtain the channel location of each channel of the multi-channel ground penetrating radar system in the single-channel data feature library;

[0025] The effective extraction module is used to calculate the retrieval channel distance matrix based on the location of each channel, and to extract effective retrieval pairs using the constructed spatial prior constraint matrix and the retrieval channel distance matrix.

[0026] The effective channel module is used to determine the effective retrieval channel from the candidate effective retrieval channel list after determining the list of candidate effective retrieval channels based on the Euclidean distance between the single channel data features in the channel dimension data features and the single channel data features in the single channel data feature library.

[0027] The positioning calculation module is used to perform linear fitting based on the effective retrieval channels, calculate the array center point position of the multi-channel ground penetrating radar system, and obtain the ground penetrating radar image retrieval positioning results.

[0028] In one embodiment, the above-mentioned ground-penetrating radar image retrieval and positioning device based on spatial prior constraints further includes:

[0029] The 3D acquisition module is used to acquire 3D map data from a multi-channel ground-penetrating radar system, resulting in 3D volume data composed of multiple single-channel 3D data.

[0030] The library construction module is used to extract features from each single-channel dimension of the three-dimensional volume data to build a single-channel data feature library for a multi-channel ground penetrating radar system.

[0031] In one embodiment, the library construction module is used to perform multi-level wavelet decomposition on each single-channel dimension data in the three-dimensional volume data from the frequency domain to obtain the single-channel data features of each single-channel dimension data, thereby obtaining a single-channel data feature library.

[0032] In one embodiment, the distance constraint between valid retrieval pairs during the process of obtaining valid retrieval pairs is:

[0033]

[0034] Where Λ represents the distance error threshold, and R ij The distance from the i-th channel to the j-th channel of the array antenna is represented by D, denoted as the spatial prior constraint matrix. ij Let represent the distance between the i-th channel and the j-th channel obtained through location retrieval, denoted as the retrieval channel distance matrix, and abs() represents calculating the absolute value.

[0035] On another front, a ground-penetrating radar positioning device is also provided, including a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it implements the steps of the above-mentioned ground-penetrating radar image retrieval and positioning method based on spatial prior constraints.

[0036] Furthermore, a computer-readable storage medium is also provided, on which a computer program is stored, which, when executed by a processor, implements the steps of the above-described ground-penetrating radar image retrieval and localization method based on spatial prior constraints.

[0037] One of the above technical solutions has the following advantages and beneficial effects:

[0038] The aforementioned ground-penetrating radar (GPR) image retrieval and positioning method and apparatus based on spatial prior constraints improve positioning accuracy by generally utilizing single-channel GPR data information for similarity measurement and combining it with spatial prior constraints of the array. Specifically, firstly, multi-level wavelet transform is used to extract features based on the characteristics of GPR A-scan data, and a single-channel GPR data feature library is constructed. Then, since the amount of GPR data information is small, the probability of mismatches is still relatively high when using only one-dimensional data for registration. Therefore, a spatial prior constraint matrix of the array channels is added to determine effective retrieval pairs using fixed distance information between channels. This reduces the impact of mismatches caused by relying solely on similarity retrieval, effectively improving the positioning accuracy of GPR in cases of incomplete maps, and providing an important guarantee for achieving safe and reliable GPR navigation and positioning. Attached Figure Description

[0039] To more clearly illustrate the technical solutions in the embodiments of this application or the conventional technology, the drawings used in the description of the embodiments or the conventional technology will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0040] Figure 1 This is a flowchart illustrating a ground-penetrating radar image retrieval and localization method based on spatial prior constraints in one embodiment.

[0041] Figure 2 This is a schematic diagram of the data format of a multi-channel ground penetrating radar in one embodiment, where (a) is single-channel 3D data, (b) is data obtained by scanning multiple times along the region of interest in a single channel, (c) is a real-time image obtained by a single acquisition by the multi-channel ground penetrating radar system, and (d) is 3D volume data.

[0042] Figure 3 This is a flowchart illustrating a ground-penetrating radar image retrieval and localization method based on spatial prior constraints in another embodiment.

[0043] Figure 4 This is a schematic diagram of the multi-level wavelet decomposition process for A-scan data in one embodiment;

[0044] Figure 5 This is a schematic diagram of the module structure of a ground-penetrating radar image retrieval and positioning device based on spatial prior constraints in one embodiment. Detailed Implementation

[0045] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0046] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to be limiting of the application.

[0047] It should be noted that, in this document, the reference to "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The presentation of this phrase in various locations throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments.

[0048] Those skilled in the art will understand that the embodiments described herein can be combined with other embodiments. The term "and / or" as used in this specification and the appended claims refers to any combination of one or more of the associated listed items, and all possible combinations thereof.

[0049] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0050] Please see Figure 1 In one embodiment, a ground-penetrating radar image retrieval and localization method based on spatial prior constraints is provided, including the following processing steps S12 to S22:

[0051] S12, acquire a pre-built single-channel data feature library for multi-channel ground penetrating radar systems.

[0052] It is understandable that the single-channel data feature library (which can be denoted as W) corresponding to the multi-channel ground penetrating radar system can be pre-collected using the multi-channel ground penetrating radar system to collect ground penetrating radar 3D map data. Then, feature extraction is performed on the 3D volume data composed of multiple A-scan data collected to obtain the low-frequency information of these A-scan data in the ground penetrating radar 3D map data, also known as low-frequency features, thereby ultimately constructing the single-channel data feature library.

[0053] It should be noted that for existing basic knowledge aspects such as ground-penetrating radar (GPR) map acquisition and storage, data can be collected using a multi-channel GPR system. This multi-channel GPR system can include existing components such as a high-precision positioning system (e.g., GPS or BeiDou), GPR, and a high-performance computer. The GPR data obtained through data acquisition is combined with high-precision positioning (e.g., GPS or BeiDou) location tags to generate a grid map. The data format of the multi-channel GPR system can be described as follows: Figure 2 As shown, x represents the travel direction of the multi-channel ground-penetrating radar system, and y represents the array channel arrangement direction of the multi-channel ground-penetrating radar system. Single-channel dimensional data is as follows: Figure 2 As shown in (a), this is called an A-scan; multiple A-scans can be obtained by scanning along the region of interest multiple times using a single channel, such as... Figure 2 As shown in (b), this is called B-scan; the real-time image obtained by a single acquisition of a multi-channel ground-penetrating radar system is as follows: Figure 2 As shown in (c), this is called D-scan, which is two-dimensional (2D) ground-penetrating radar data; the corresponding real-time images obtained by multi-channel acquisition along the region of interest can be shown as follows: Figure 2 As shown in (d), this is called three-dimensional (3D) volume data.

[0054] S14 utilizes a multi-channel ground-penetrating radar system to acquire real-time two-dimensional ground-penetrating radar data, and uses multi-level wavelet transform to extract channel dimension data features from the two-dimensional ground-penetrating radar data.

[0055] It is understandable that a multi-channel ground-penetrating radar system can be used to obtain real-time two-dimensional ground-penetrating radar data, denoted as I = [A1, A2, ..., A...]. m ], where m represents the number of channels in the array, and m A-scan data A are obtained according to the array arrangement. i (i = 1, 2, ..., m), the channel dimension data features of real-time ground-penetrating radar two-dimensional data can be obtained using multi-level wavelet transform, denoted as... in, This represents the channel-dimensional data feature corresponding to channel i of the array. The specific decomposition process of the multi-level wavelet transform will be given below.

[0056] S16. Based on the channel dimension data characteristics, a similarity search is performed in the single-channel data feature library to obtain the channel position of each channel of the multi-channel ground penetrating radar system in the single-channel data feature library.

[0057] It is understandable that, specifically, for the single-channel data feature retrieval in this step, the channel-dimensional data features obtained in the previous step will be... Divided into m one-dimensional single-channel data features Then, similarity searches are performed sequentially in the single-channel data feature library W. Euclidean distance can be used as the similarity measure of features, as shown in formula (1), where, This represents the single-channel data feature to be queried. Correspondingly, f can be used to represent the single-channel data feature in the single-channel data feature library W, where M represents the Euclidean distance between two single-channel data features. The smaller the value of M, the smaller the distance between the two single-channel data features, indicating that the two single-channel data features are more similar. The single-channel data feature corresponding to the minimum value of M is selected. The position is used as the position of the feature to be queried, and finally the position p of each channel in the single-channel data feature library W is obtained. i Thus, the set of channel positions P(C) of the real-time image (corresponding to the real-time acquisition of ground-penetrating radar two-dimensional data) can be obtained.

[0058]

[0059] S18. Calculate the retrieval channel distance matrix based on the location of each channel, and extract effective retrieval pairs using the constructed spatial prior constraint matrix and the retrieval channel distance matrix to obtain effective retrieval pairs.

[0060] This step involves extracting effective retrieval pairs based on spatial prior constraints. Specifically, it utilizes existing spatial prior information (such as fixed distance information between channels) from the array channels themselves to construct a spatial prior constraint matrix R. Based on the channel positions obtained in the previous step, it calculates the distance information between channels to obtain the retrieval channel distance matrix D. The spatial prior constraint matrix R is determined by the fixed distance between the channels of the multi-channel ground-penetrating radar system's antenna array, with a channel spacing of d. c The number of channels is m, and the spatial prior constraint matrix R is shown in formula (2). The retrieval channel distance matrix D corresponding to the retrieval results is determined by the position p of each channel. i The decision is as shown in formula (3).

[0061]

[0062]

[0063] The distance between valid search pairs should satisfy the constraint, as shown in formula (4):

[0064]

[0065] Where Λ is the acceptable distance error threshold, used to balance the recall and precision of the retrieval. R ij The distance from the i-th channel to the j-th channel of the array antenna is represented by D, denoted as the spatial prior constraint matrix. ij Let represent the distance between the i-th channel and the j-th channel obtained through location retrieval, denoted as the retrieval channel distance matrix, and abs() denotes the absolute value. The effective retrieval pair can be calculated according to formula (4), denoted as (u t ,v t ), where u t ,v t Let t represent the array channel number pair corresponding to the t-th valid retrieval pair, where t = 1, 2, ..., s, and s represents the total number of valid retrieval pairs obtained.

[0066] S20. After determining the list of candidate valid retrieval channels based on the valid retrieval pairs, the valid retrieval channels are determined from the list of candidate valid retrieval channels based on the Euclidean distance between the single-channel data features in the channel dimension data features and the single-channel data features in the single-channel data feature library.

[0067] Specifically, based on the valid search pairs obtained above, a list of candidate valid search channels is obtained: list = [C1, C2, ..., C...]. k], where k represents the total number of candidate valid retrieval channels C contained in the valid retrieval pair. Based on the feature distance M in step S16, the retrieval results M(C) of the candidate valid channels are obtained. M(C) are sorted in ascending order, and the top N candidate valid channels are determined as valid retrieval channels to obtain the valid retrieval channel list list2. The determination of N can be made by formula (5):

[0068]

[0069] S22, perform linear fitting based on the effective retrieval channels, calculate the array center point position of the multi-channel ground penetrating radar system, and obtain the ground penetrating radar image retrieval and positioning results.

[0070] It can be understood that, based on the N valid retrieval channels obtained in step S20, the coordinates of N given points [(x1,y1),(x2,y2),...,(x...]] can be determined. N ,y N ]], take any two points and calculate the array distribution line l: x=ay+b, to obtain the slope matrix A:

[0071]

[0072] Where, slope a ij =(x i -x j ) / (y i -y j ).

[0073] The slope value that appears most frequently in the slope matrix A is the slope a corresponding to the array distribution line l. Based on the slope matrix A and the slope a, a pair of points (i,j) is selected to determine this slope. Then, based on the given point coordinates (x...j ... i ,y i Determine the line intercept b = x1 - ay1.

[0074] Based on the array distribution line l and the given point coordinates (x) i ,y i ), and the coordinates of the given point (x i ,y i For the corresponding channel number C, calculate the array center point position (px, py):

[0075]

[0076] At this point, the position of the array center point (px, py) is the final ground-penetrating radar image retrieval and positioning result.

[0077] The aforementioned ground-penetrating radar (GPR) image retrieval and localization method based on spatial prior constraints improves localization accuracy by utilizing single-channel GPR data for similarity measurement and combining it with spatial prior constraints of the array. Specifically, firstly, multi-level wavelet transform is used to extract features based on the characteristics of GPR A-scan data, and a single-channel GPR data feature library is constructed. Then, since the amount of GPR data is limited, the probability of mismatches is still relatively high when using only one-dimensional data for registration. Therefore, a spatial prior constraint matrix of the array channels is added to determine effective retrieval pairs using fixed distance information between channels. This reduces the impact of mismatches caused by relying solely on similarity retrieval, effectively improving the localization accuracy of GPR in cases of incomplete maps, i.e., low-overlapping images, and providing an important guarantee for achieving safe and reliable GPR navigation and localization.

[0078] In one embodiment, the above-described ground-penetrating radar image retrieval and localization method based on spatial prior constraints further includes the following steps:

[0079] A multi-channel ground-penetrating radar system was used to acquire three-dimensional map data from ground-penetrating radar, resulting in three-dimensional volume data composed of multiple single-channel dimensional data.

[0080] Feature extraction is performed on each single-channel dimension data in the three-dimensional volume data to construct a single-channel data feature library for a multi-channel ground penetrating radar system.

[0081] It is understood that, in this embodiment, ground-penetrating radar (GPR) system can also be used directly in the current application scenario to collect GPR 3D map data on-site, obtaining 3D volume data V composed of multiple A-scan data. Feature extraction is performed on each A-scan data in V to construct a single-channel GPR data feature library. Specifically, considering that the low-frequency information of GPR data is relatively stable, the signal is decomposed in the frequency domain, and wavelet decomposition is used to obtain the low-frequency features of the GPR data. Simultaneously, since the dB wavelet family is quite similar to the GPR signal, dB wavelets are used to decompose the one-dimensional channel data, thereby achieving efficient single-channel data feature library construction.

[0082] In one embodiment, the step of extracting features from each single-channel dimension of the three-dimensional volume data to construct a single-channel data feature library for a multi-channel ground-penetrating radar system may further include the following processing:

[0083] Multi-level wavelet decomposition is performed on each single-channel dimension data in the three-dimensional volume data in the frequency domain to obtain the single-channel data features of each single-channel dimension data, and a single-channel data feature library is obtained.

[0084] Understandable, such as Figure 3As shown, in this embodiment, in the frequency domain, multi-level wavelet transform can be used to decompose the single-channel dimensional data corresponding to each A-scan data in the 3D volume data. In each level of wavelet transform, the input signal y[n] (i.e., A-scan data) is convolved with the high-pass filter h[n] and the low-pass filter l[n] in the system, and downsampled respectively to generate detail coefficients E1 and approximation coefficients E2. In the multi-level wavelet transform, the approximation coefficients E2 are provided to the next level, and the same process is repeated to obtain all the single-channel data features. Finally, a ground-penetrating radar single-channel data feature library W is constructed, and its decomposition process is as follows: Figure 4 As shown, h1, h2, h3… are all high-pass filters, and l1, l2, l3… are all low-pass filters. The specific number of filters can be determined based on the number of stages in the wavelet transform. Through the above multi-stage wavelet transform processing, more efficient and accurate single-channel data feature extraction can be achieved.

[0085] In one embodiment, the distance constraint between valid search pairs during the process of obtaining valid search pairs is:

[0086]

[0087] Where Λ represents the distance error threshold, and R ij The distance from the i-th channel to the j-th channel of the array antenna is represented by D, denoted as the spatial prior constraint matrix. ij Let represent the distance between the i-th channel and the j-th channel obtained through location retrieval, denoted as the retrieval channel distance matrix, and abs() represents calculating the absolute value.

[0088] It is understandable that, since the spatial prior constraint matrix R and the retrieval channel distance matrix D are symmetric matrices, formula (4) is adjusted to formula (8) to reduce data duplication in the effective retrieval pair records. Therefore, in this embodiment, the effective retrieval pair pairs = [(u1,v1),(u2,v2),...,(u...] can be directly obtained according to formula (8). s ,v s )], where s represents the total number of valid search pairs obtained, which is more computationally efficient and has less resource overhead.

[0089] It should be understood that, although Figure 1 , Figure 3 and Figure 4 The steps in the flowchart are shown sequentially as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order requirement for the execution of these steps; they can be executed in other orders. Figure 1 , Figure 3 and Figure 4At least some of the steps may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least some of the sub-steps or stages of other steps.

[0090] Please see Figure 5 In one embodiment, a ground-penetrating radar (GPR) image retrieval and positioning device 100 based on spatial prior constraints is provided, including a feature library acquisition module 11, a real-time extraction module 13, a channel position module 15, an effective extraction module 17, an effective channel module 19, and a positioning calculation module 21. The feature library acquisition module 11 acquires a pre-constructed single-channel data feature library for a multi-channel GPR system. The real-time extraction module 13 uses real-time GPR two-dimensional data acquired by the multi-channel GPR system and extracts channel-dimensional data features from the GPR two-dimensional data using multi-level wavelet transform. The channel position module 15 performs a similarity search in the single-channel data feature library based on the channel-dimensional data features to obtain the channel position of each channel of the multi-channel GPR system in the single-channel data feature library. The effective extraction module 17 calculates the retrieval channel distance matrix based on the channel position and extracts effective retrieval pairs using the constructed spatial prior constraint matrix and the retrieval channel distance matrix. The effective channel module 19 is used to determine the effective retrieval channel list after determining the candidate effective retrieval channel list based on the effective retrieval pairs, and then determine the effective retrieval channel from the candidate effective retrieval channel list based on the Euclidean distance between the single-channel data features in the channel dimension data features and the single-channel data features in the single-channel data feature library. The positioning calculation module 21 is used to perform straight line fitting based on the effective retrieval channels, calculate the array center point position of the multi-channel ground penetrating radar system, and obtain the ground penetrating radar image retrieval positioning result.

[0091] The aforementioned ground-penetrating radar (GPR) image retrieval and positioning device 100 based on spatial prior constraints improves positioning accuracy by generally utilizing single-channel GPR data information for similarity measurement and combining it with spatial prior constraints of the array. Specifically, firstly, it extracts features using multi-level wavelet transform based on the characteristics of GPR A-scan data and constructs a single-channel GPR data feature library. Then, since the amount of GPR data information is small, the probability of mismatch is still relatively high when using only one-dimensional data for registration. Therefore, a spatial prior constraint matrix of the array channels is added to determine effective retrieval pairs using fixed distance information between channels. This reduces the impact of mismatch caused by relying solely on similarity retrieval, effectively improving the positioning accuracy of GPR in cases of incomplete maps, i.e., low-overlapping images, and providing an important guarantee for achieving safe and reliable GPR navigation and positioning.

[0092] In one embodiment, the aforementioned ground-penetrating radar image retrieval and positioning device 100 based on spatial prior constraints further includes a 3D acquisition module and a library construction module. The 3D acquisition module is used to acquire 3D map data from a multi-channel ground-penetrating radar system, obtaining 3D volume data composed of multiple single-channel dimensional data. The library construction module is used to extract features from each single-channel dimensional data in the 3D volume data, constructing a single-channel data feature library for the multi-channel ground-penetrating radar system.

[0093] In one embodiment, when constructing a single-channel data feature library, the library construction module can specifically be used to perform multi-level wavelet decomposition on each single-channel dimension data in the three-dimensional volume data from the frequency domain to obtain the single-channel data features of each single-channel dimension data, thereby obtaining the single-channel data feature library.

[0094] In one embodiment, the distance constraint between valid search pairs during the process of obtaining valid search pairs is:

[0095]

[0096] Where Λ represents the distance error threshold, and R ij The distance from the i-th channel to the j-th channel of the array antenna is represented by D, denoted as the spatial prior constraint matrix. ij Let represent the distance between the i-th channel and the j-th channel obtained through location retrieval, denoted as the retrieval channel distance matrix, and abs() represents calculating the absolute value.

[0097] For specific limitations regarding the ground-penetrating radar image retrieval and positioning device 100 based on spatial prior constraints, please refer to the corresponding limitations of the ground-penetrating radar image retrieval and positioning method based on spatial prior constraints mentioned above, which will not be repeated here. Each module in the aforementioned ground-penetrating radar image retrieval and positioning device 100 based on spatial prior constraints can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in hardware or independently of a device with data processing capabilities, or stored in software in the memory of the aforementioned device, so that the processor can call and execute the operations corresponding to each module. The aforementioned device can be, but is not limited to, various types of radar data calculation and processing devices already existing in the art.

[0098] In one embodiment, a ground-penetrating radar (GPR) positioning device is also provided, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to perform the following processing steps: acquiring a pre-constructed single-channel data feature library for a multi-channel GPR system; acquiring real-time GPR two-dimensional data using the multi-channel GPR system, and extracting channel-dimensional data features from the GPR two-dimensional data using multi-level wavelet transform; performing a similarity search in the single-channel data feature library based on the channel-dimensional data features to obtain the channel positions of each channel of the multi-channel GPR system in the single-channel data feature library; calculating the retrieval channel distance matrix based on the channel positions, and extracting effective retrieval pairs using the constructed spatial prior constraint matrix and the retrieval channel distance matrix to obtain effective retrieval pairs; determining a candidate list of effective retrieval channels based on the effective retrieval pairs, and determining effective retrieval channels from the candidate list of effective retrieval channels based on the Euclidean distance between the single-channel data features in the channel-dimensional data features and the single-channel data features in the single-channel data feature library; performing linear fitting based on the effective retrieval channels to calculate the array center point position of the multi-channel GPR system, and obtaining the GPR image retrieval and positioning result.

[0099] It is understood that, in addition to the memory and processor mentioned above, the ground-penetrating radar positioning device may also include other hardware and software components not listed in this specification. The specific components can be determined according to the model of the positioning device in different application scenarios, and will not be listed and described in detail in this specification.

[0100] In one embodiment, when the processor executes the computer program, it can also implement the steps or sub-steps added in the various embodiments of the above-described ground-penetrating radar image retrieval and positioning method based on spatial prior constraints.

[0101] In one embodiment, a computer-readable storage medium is also provided, on which a computer program is stored. When executed by a processor, the computer program performs the following processing steps: acquiring a pre-constructed single-channel data feature library for a multi-channel ground-penetrating radar system; acquiring real-time two-dimensional ground-penetrating radar data using the multi-channel ground-penetrating radar system, and extracting channel-dimensional data features from the two-dimensional ground-penetrating radar data using multi-level wavelet transform; performing a similarity search in the single-channel data feature library based on the channel-dimensional data features to obtain the channel positions of each channel of the multi-channel ground-penetrating radar system in the single-channel data feature library; calculating the retrieval channel distance matrix based on the channel positions, and extracting effective retrieval pairs using the constructed spatial prior constraint matrix and the retrieval channel distance matrix to obtain effective retrieval pairs; determining a candidate list of effective retrieval channels based on the effective retrieval pairs, and determining effective retrieval channels from the candidate list of effective retrieval channels based on the Euclidean distance between the single-channel data features in the channel-dimensional data features and the single-channel data features in the single-channel data feature library; performing linear fitting based on the effective retrieval channels to calculate the array center point position of the multi-channel ground-penetrating radar system, and obtaining the ground-penetrating radar image retrieval and positioning results.

[0102] In one embodiment, when the computer program is executed by the processor, it can also implement the steps or sub-steps added to the various embodiments of the ground-penetrating radar image retrieval and positioning method based on spatial prior constraints.

[0103] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), erasable programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), memory bus DRAM (RDRAM), and interface DRAM (DRDRAM), etc.

[0104] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0105] The above embodiments merely illustrate several implementation methods of this application, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, all of which fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A ground-penetrating radar image retrieval and localization method based on spatial prior constraints, characterized in that, Including the following steps: Acquire a pre-built single-channel data feature library for multi-channel ground-penetrating radar systems; The multi-channel ground-penetrating radar system is used to collect real-time two-dimensional ground-penetrating radar data, and multi-level wavelet transform is used to extract channel dimension data features from the two-dimensional ground-penetrating radar data. Based on the channel dimension data features, a similarity search is performed in the single-channel data feature library to obtain the channel position of each channel of the multi-channel ground penetrating radar system in the single-channel data feature library; Calculate the retrieval channel distance matrix based on the location of each channel, and extract effective retrieval pairs using the constructed spatial prior constraint matrix and the retrieval channel distance matrix to obtain effective retrieval pairs. After determining the candidate effective retrieval channel list based on the effective retrieval pair, the effective retrieval channel is determined from the candidate effective retrieval channel list based on the Euclidean distance between the single-channel data features in the channel dimension data features and the single-channel data features in the single-channel data feature library. Based on the effective retrieval channels, a straight line fit is performed to calculate the position of the array center point of the multi-channel ground penetrating radar system, thereby obtaining the ground penetrating radar image retrieval and positioning results.

2. The ground-penetrating radar image retrieval and localization method based on spatial prior constraints according to claim 1, characterized in that, It also includes the following steps: The multi-channel ground-penetrating radar system is used to acquire three-dimensional map data of ground-penetrating radar, and three-dimensional volume data composed of multiple single-channel dimensional data is obtained. Feature extraction is performed on each single-channel dimension data in the three-dimensional volume data to construct the single-channel data feature library of the multi-channel ground penetrating radar system.

3. The ground-penetrating radar image retrieval and localization method based on spatial prior constraints according to claim 2, characterized in that, The step of extracting features from each single-channel dimension of the three-dimensional volume data to construct the single-channel data feature library of the multi-channel ground-penetrating radar system includes: Multi-level wavelet decomposition is performed on each single-channel dimension data in the three-dimensional volume data in the frequency domain to obtain the single-channel data features of each single-channel dimension data, thereby obtaining the single-channel data feature library.

4. The ground-penetrating radar image retrieval and localization method based on spatial prior constraints according to any one of claims 1 to 3, characterized in that, In the process of obtaining valid search pairs, the distance constraint between valid search pairs is: Where Λ represents the distance error threshold, and R ij The distance from the i-th channel to the j-th channel of the array antenna is represented by D, denoted as the spatial prior constraint matrix. ij Let represent the distance between the i-th channel and the j-th channel obtained through location retrieval, denoted as the retrieval channel distance matrix, and abs() represents calculating the absolute value.

5. A ground-penetrating radar image retrieval and positioning device based on spatial prior constraints, characterized in that, include: The feature library acquisition module is used to acquire a pre-built single-channel data feature library for multi-channel ground penetrating radar systems; The real-time extraction module is used to collect real-time two-dimensional ground-penetrating radar data using the multi-channel ground-penetrating radar system, and to extract channel dimension data features from the two-dimensional ground-penetrating radar data using multi-level wavelet transform. The channel location module is used to perform a similarity search in the single-channel data feature library based on the channel dimension data features to obtain the channel location of each channel of the multi-channel ground penetrating radar system in the single-channel data feature library. The effective extraction module is used to calculate the retrieval channel distance matrix based on the location of each channel, and to extract effective retrieval pairs using the constructed spatial prior constraint matrix and the retrieval channel distance matrix to obtain effective retrieval pairs. The effective channel module is used to determine the effective retrieval channel from the candidate effective retrieval channel list after determining the list of candidate effective retrieval channels based on the Euclidean distance between the single-channel data features in the channel dimension data features and the single-channel data features in the single-channel data feature library. The positioning calculation module is used to perform linear fitting based on the effective retrieval channels, calculate the position of the array center point of the multi-channel ground penetrating radar system, and obtain the ground penetrating radar image retrieval positioning result.

6. The ground-penetrating radar image retrieval and positioning device based on spatial prior constraints according to claim 5, characterized in that, Also includes: The three-dimensional acquisition module is used to acquire three-dimensional map data of the ground penetrating radar system using the multi-channel ground penetrating radar system, and obtain three-dimensional volume data composed of multiple single-channel three-dimensional data. The library construction module is used to extract features from each single-channel dimension data in the three-dimensional volume data to construct the single-channel data feature library of the multi-channel ground penetrating radar system.

7. The ground-penetrating radar image retrieval and positioning device based on spatial prior constraints according to claim 6, characterized in that, The library construction module is used to perform multi-level wavelet decomposition on each single-channel dimension data in the three-dimensional volume data from the frequency domain to obtain the single-channel data features of each single-channel dimension data, and obtain the single-channel data feature library.

8. The ground-penetrating radar image retrieval and positioning device based on spatial prior constraints according to any one of claims 5 to 7, characterized in that, In the process of obtaining valid search pairs, the distance constraint between valid search pairs is: Where Λ represents the distance error threshold, and R ij The distance from the i-th channel to the j-th channel of the array antenna is represented by D, denoted as the spatial prior constraint matrix. ij Let represent the distance between the i-th channel and the j-th channel obtained through location retrieval, denoted as the retrieval channel distance matrix, and abs() represents calculating the absolute value.

9. A ground-penetrating radar positioning device, comprising a memory and a processor, characterized in that, The memory stores a computer program, and when the processor executes the computer program, it implements the steps of the ground-penetrating radar image retrieval and positioning method based on spatial prior constraints as described in any one of claims 1 to 4.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the ground-penetrating radar image retrieval and positioning method based on spatial prior constraints as described in any one of claims 1 to 4.

Images