Target detection method, system and device based on ground penetrating radar and storage medium
By normalizing and adjusting the gray levels of ground-penetrating radar echo data, and combining this with the CLAHE method for local adaptive gray-level optimization, the problem of poor ground-penetrating radar image quality was solved, the target detection accuracy of the YOLO model was improved, and the false negative rate was reduced.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- HUNAN ECONOVEL TECH CO LTD
- Filing Date
- 2026-01-22
- Publication Date
- 2026-05-08
AI Technical Summary
Ground penetrating radar (Bscan) images have low signal-to-noise ratios and weak target features, resulting in poor image quality. Existing methods struggle to effectively suppress noise, and the YOLO model exhibits low detection accuracy and high false negative rate in the field of ground penetrating radar.
By normalizing and grayscale processing the ground-penetrating radar echo data, performing grayscale level adjustment and contrast enhancement in blocks, and using the CLAHE method to perform adaptive grayscale distribution optimization in local areas, the data is then input into a pre-trained YOLO model for target detection.
It improves the accuracy of target detection, reduces the false negative rate, makes it easier for the YOLO model to separate target features from the background, enhances image contrast and clarity, and reduces false negatives caused by depth differences.
Smart Images
Figure CN121995374A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of target detection, specifically relating to a target detection method, system, device, and storage medium based on ground-penetrating radar. Background Technology
[0002] Ground-penetrating radar (GPR) is widely used in urban road defect detection (such as cavity identification), underground pipeline detection, and archaeological exploration. Its core function is to locate targets by analyzing the hyperbolic characteristics of target echoes (such as metal pipes, cracks, and cavities) in Bscan echo images.
[0003] However, issues such as low signal-to-noise ratio, weak target features, and poor image quality result in insufficient contrast in Bscan images, making it difficult to highlight hyperbolic features. Existing ground-penetrating radar (GPR) enhancement methods, such as linear upscaling, wavelet denoising, and global histogram homogenization, struggle to effectively suppress noise while preserving weak signals, leading to a high false negative rate in subsequent target detection. In terms of detection, while the YOLO series of deep learning-based models have performed excellently in general target detection in recent years, they still face challenges in the GPR field: firstly, the small size of GPR target echoes makes it difficult for YOLO's default anchor boxes to match; secondly, the high noise interference and low contrast of raw Bscan images make direct input into the model prone to false detections. Summary of the Invention
[0004] The technical problem to be solved by the present invention is to provide a target detection method, system, device and storage medium based on ground penetrating radar, which improves the target detection accuracy and reduces the false detection rate by contrast enhancement of images.
[0005] A target detection method based on ground-penetrating radar includes: The ground-penetrating radar travels in a straight line and detects the target during its operation, acquiring echo data of the target at different detection points. The echo data is normalized to obtain a normalized image, and the normalized image is converted into a grayscale image to obtain a grayscale image. The grayscale image is divided into several grayscale sub-images; Based on the grayscale subimage, the frequency of occurrence of each grayscale level is obtained, and the cumulative distribution value is calculated based on the frequency of occurrence of each grayscale level. If the cumulative distribution value is greater than a preset threshold, the grayscale sub-image is cropped, and the grayscale levels with cumulative distribution values greater than the preset threshold are adjusted and evenly distributed to other grayscale levels. The cumulative distribution value of the adjusted grayscale sub-image is calculated until there are no cumulative distribution values greater than the preset threshold. The cumulative distribution values of the adjusted grayscale sub-image are normalized to obtain the normalized value; Mapping is performed based on the normalization value to enhance the contrast of the grayscale image, resulting in an enhanced image. The enhanced image is then resized and pixel-adjusted to obtain the adjusted image; The adjusted image is input into a pre-trained YOLO model to obtain the target type, location, and confidence level of the target to be detected, and the target type, location, and confidence level are used as the detection result.
[0006] Optionally, the step of normalizing the echo data to obtain normalized data, and converting the normalized data into a grayscale image, includes: The echo data is represented as a radar echo signal, as follows:
[0007] in, For the first The coordinates of each detection point in the detection plane The total number of detection points. For time, and The ground-penetrating radar at the 1st The amplitude and phase of the echo received by each detection point; A two-dimensional image is drawn based on the radar echo signal; The two-dimensional image is normalized to obtain a normalized image; The normalized image is converted into a grayscale image to obtain a grayscale image.
[0008] Optionally, the step of obtaining the frequency of occurrence of each gray level based on the gray-scale sub-image and calculating the cumulative distribution value based on the frequency of occurrence of each gray level includes: Based on the grayscale sub-image, count the number of occurrences of each grayscale level; The cumulative distribution value is obtained based on the number of times the gray level appears and the total number of pixels in the gray sub-image.
[0009] Optionally, the calculated cumulative distribution value of the adjusted grayscale sub-image includes: The number of remaining pixels after cropping is calculated based on the occurrence frequency and the cropping threshold. The number of occurrences after cropping is calculated based on the remaining number of pixels and the cropping threshold. The cumulative distribution value of the adjusted grayscale sub-image is calculated based on the number of occurrences after cropping.
[0010] Optionally, the cumulative distribution value of the adjusted grayscale subimage is normalized to obtain a normalized value, which is expressed as:
[0011] in, , This represents the cumulative distribution value of the adjusted grayscale subimage. It is the normalized value.
[0012] Optionally, inputting the adjusted image into a pre-trained YOLO model to obtain the target type, location, and confidence level of the target to be detected includes: Extract the features to be predicted from the adjusted image; The features to be predicted are input into a pre-trained YOLO model to obtain the target type, location, and confidence level of the target to be detected.
[0013] A ground-penetrating radar-based target detection system includes: The data acquisition module is used to use ground penetrating radar to move along a straight line, detect the target to be detected during the operation, and acquire the echo data of the target to be detected at different detection points; The image processing module is used to normalize the echo data to obtain a normalized image, and convert the normalized image into a grayscale image to obtain a grayscale image. A splitting module is used to divide the grayscale image into several grayscale sub-images; The calculation module is used to obtain the frequency of occurrence of each gray level based on the gray level sub-image, and to obtain the cumulative distribution value based on the frequency of occurrence of the gray level; The adjustment module is used to crop the grayscale sub-image if the cumulative distribution value is greater than a preset threshold, adjust the grayscale levels with cumulative distribution values greater than the preset threshold, and then evenly distribute them to other grayscale levels, and calculate the cumulative distribution value of the adjusted grayscale sub-image until there are no cumulative distribution values greater than the preset threshold. The normalization module is used to normalize the cumulative distribution values of the adjusted grayscale sub-image to obtain a normalized value. The enhancement module is used to perform mapping based on the normalization value to achieve contrast enhancement of the grayscale image, thereby obtaining an enhanced image; The image adjustment module is used to adjust the size and pixels of the enhanced image to obtain an adjusted image; The detection module is used to input the adjusted image into a pre-trained YOLO model to obtain the target type, location, and confidence level of the target to be detected, and to use the target type, location, and confidence level as the detection result.
[0014] The computing module includes: The statistics unit is used to count the number of occurrences of each gray level based on the grayscale sub-image; The calculation unit is used to obtain the cumulative distribution value based on the number of occurrences of the gray level and the total number of pixels in the gray sub-image.
[0015] A terminal device includes a memory and a processor. The memory stores a computer program that can run on the processor. When the processor loads and executes the computer program, it employs a target detection method based on ground-penetrating radar.
[0016] A computer-readable storage medium storing a computer program, which, when loaded and executed by a processor, employs a ground-penetrating radar-based target detection method.
[0017] The beneficial effects of this invention are: Ground-penetrating radar (GPR) operates along a straight line, detecting targets during its operation and acquiring echo data of the detected targets at different detection points. The echo data is normalized to obtain a normalized image, which is then converted to a grayscale image. This grayscale image is divided into several grayscale sub-images, and a histogram of grayscale values is calculated for each sub-image. Contrast enhancement is performed on the grayscale image based on the histograms to obtain an enhanced image. The enhanced image is then resized and pixel-adjusted to obtain an adjusted image. This adjusted image is input into a pre-trained YOLO model to obtain the target type, location, and confidence score of the detected target. The target type, location, and confidence score are then used as the detection result. Compared to traditional processing methods, this application, by performing independent histogram equalization on image blocks, can specifically enhance the contrast of each small region, making previously blurred hyperbolic edges and vertices clearer and more continuous. This allows the YOLO model to more easily separate the target from the background, thereby improving detection accuracy. Secondly, if traditional global histogram equalization is used, the grayscale distribution of the entire image is forcibly stretched, which may lead to amplified background noise and even overexposure and loss of detail in the target area. CLAHE's local adaptive feature optimizes contrast only within the current image patch, highlighting the target while suppressing excessive changes in non-critical areas, thus preserving more useful information. Finally, Bscan images exhibit significant differences in echo intensity at different depths (time axis), and traditional normalization may not be able to account for both shallow and deep targets. CLAHE adaptively adjusts the grayscale distribution in different local regions, essentially performing adaptive brightness and contrast correction at different depths, ensuring that targets at both shallow and deep depths are visually presented in a balanced manner, reducing false negatives due to depth. This solves the problem of low local contrast in ground-penetrating radar Bscan images, improving the accuracy of target detection and reducing the false negative rate. Attached Figure Description
[0018] Figure 1 This is a schematic flowchart of a target detection method based on ground-penetrating radar according to the present invention; Figure 2 This is a comparison chart of the detection results of the detection method of the present invention and the traditional detection method. Detailed Implementation
[0019] A target detection method based on ground-penetrating radar, such as Figure 1 As shown, it includes: S1. Use ground penetrating radar to run in a straight line, and detect the target to be detected during the operation, and obtain the echo data of the target to be detected at different detection points; Specifically, a traditional ground-penetrating radar is used to move along a straight line and detect the target during the movement, thereby obtaining the echo data of the target at different detection points on the straight line, namely the Bscan echo data of the ground-penetrating radar. Based on this data, a two-dimensional Bscan image can be obtained. The image is then normalized and grayscaled to obtain the preliminary Bscan grayscale.
[0020] S2. Normalize the echo data to obtain a normalized image, and convert the normalized image into a grayscale image to obtain a grayscale image. The echo data is normalized to obtain normalized data, and the normalized data is converted into a grayscale image, including: Representing the echo data as radar echo signals, as follows:
[0021] in, For the first The coordinates of each detection point in the detection plane The total number of detection points. For time, and The ground-penetrating radar at the 1st The amplitude and phase of the echo received by each detection point; from the radar echo signal, it can be seen that both amplitude and phase are quantities that change with the detection position and time. Ground-penetrating radars typically move along a straight line, therefore the coordinates of the detection points are also along a straight line. Without loss of generality, assume that the coordinate values during the movement... Monotonically changing, while coordinate values It is always 0, therefore it can be ignored. Treating radar echoes as x and t A function relating two variables is represented as:
[0022] Two-dimensional images are drawn based on radar echo signals; Specifically, both the amplitude and phase of the radar echo change with time. Fluctuations; secondly, the amplitude and phase of the radar echo change with the distance traveled. Changes. Therefore, the echo received by the radar can be moved along a straight line. Draw it as a two-dimensional image.
[0023] Normalize the two-dimensional image to obtain a normalized image; Convert the normalized image to a grayscale image to obtain a grayscale image.
[0024] Specifically, the normalization process is as follows:
[0025] in, It is a two-dimensional image. The maximum amplitude of the received signal is multiplied by 255, which means converting the received signal into a (0,255) grayscale image.
[0026] S3. Divide the grayscale image into several grayscale sub-images; Specifically, the grayscale image is divided into several grayscale sub-images, represented as follows:
[0027] in, The first subgraph obtained after partitioning Sub-images, , For the first The movement distance range corresponding to each subgraph Time range .
[0028] S4. Based on the grayscale sub-image, obtain the frequency of occurrence of each grayscale level, and calculate the cumulative distribution value based on the frequency of occurrence of each grayscale level. S5. If the cumulative distribution value is greater than the preset threshold, the grayscale sub-image is cropped, and the grayscale levels with cumulative distribution values greater than the preset threshold are adjusted and evenly distributed to other grayscale levels. The cumulative distribution value of the adjusted grayscale sub-image is calculated until there are no cumulative distribution values greater than the preset threshold. Specifically, the histogram is cropped to limit the maximum number of occurrences of each gray level, and pixels exceeding this limit are redistributed. Assume the maximum allowed occurrence threshold is T, typically three times the histogram mean. Pixels exceeding T are collected and evenly distributed among other gray levels (changing the gray values of those exceeding the threshold to other gray levels). A gray level is the total number / range of brightness levels, a set concept, while a gray value is the specific brightness value of a single pixel; the gray level defines the range of gray values.
[0029] S6. Normalize the cumulative distribution value of the adjusted grayscale sub-image to obtain the normalized value; S7. Mapping is performed based on the normalization value to enhance the contrast of the grayscale image, resulting in an enhanced image.
[0030] Based on the grayscale subimage, the frequency of occurrence of each grayscale level is obtained, resulting in the grayscale subimage and cumulative distribution value, including: Based on the grayscale sub-image, count the number of occurrences of each grayscale level; The cumulative distribution value is obtained based on the number of occurrences and the total number of pixels in the grayscale sub-image.
[0031] Specifically, a grayscale histogram is calculated for each sub-image, that is, the grayscale levels of each pixel in each sub-image are counted. Number of occurrences L=256.
[0032] The cumulative distribution value is expressed as:
[0033] Where MN is the total number of pixels in the sub-image. Let j be the number of times gray level j appears. , where i is the gray level and j is the index.
[0034] The calculation of the cumulative distribution value of the adjusted grayscale subimage includes: The number of remaining pixels after cropping is calculated based on the occurrence frequency and the cropping threshold. Specifically, the number of remaining pixels after cropping is represented as follows:
[0035] in, T represents the number of remaining pixels after cropping, and N is the sequence length of the sub-image. If the CDF of any gray level still exceeds the threshold after allocation, this step needs to be iterated until the CDF of all gray levels is less than the threshold T.
[0036] The number of occurrences after cropping is calculated based on the number of remaining pixels and the cropping threshold. The cumulative distribution value of the adjusted grayscale sub-image is calculated based on the number of times it appears after cropping.
[0037] Specifically, the CDF is recalculated based on the number of occurrences of the cropped grayscale levels, expressed as:
[0038] in, This represents the number of times the image appears after cropping.
[0039] The cumulative distribution values of the adjusted grayscale subimage are normalized to obtain the normalized value, which is represented as:
[0040] in, , This represents the cumulative distribution value of the adjusted grayscale subimage. It is the normalized value.
[0041] Then, the gray values of the original grayscale subimage are mapped to obtain new gray values, represented as:
[0042] The enhanced image is constructed using the adjusted new grayscale values.
[0043] S8. Adjust the size and pixels of the enhanced image to obtain the adjusted image; Specifically, the input image is cropped and scaled to a uniform size (e.g., 640×640 pixels), and the pixel values are normalized to [ ]. 1,1], ,in, This is the pixel value matrix of the original input image (enhanced image). This is the normalized pixel value matrix.
[0044] S9. Input the adjusted image into the pre-trained YOLO model to obtain the target type, location, and confidence level of the target to be detected, and use the target type, location, and confidence level as the detection result.
[0045] The adjusted image is input into the pre-trained YOLO model to obtain the target type, location, and confidence score of the target to be detected, including: Extract the features to be predicted from the adjusted image; The features to be predicted are input into the pre-trained YOLO model to obtain the target type, location, and confidence level of the target to be detected.
[0046] Specifically, the features to be predicted include center coordinates, width, and height. A hyperbola is extracted from the adjusted image. The center coordinates are the x and y coordinates of the maximum value of the hyperbola, the width is the length of the two endpoints of the hyperbola, and the height is the difference between the maximum and minimum values of the hyperbola (the value on the y-axis). The predicted center coordinates are: ,in, For activation function, and This is the network's raw output for the center point coordinate offset. and The predicted center point is located at the top-left corner of the grid cell (grid index multiplied by grid size), and x and y are the predicted center point coordinates of the bounding box. Width and height predictions are represented as follows: Where w and h are the width and height of the predicted bounding box, and This is the original output of the network with the bounding box width and height scaled. and Define the baseline width and height of the predefined anchor frame.
[0047] After extracting the predictive features, the predictive features are input into the pre-trained YOLO model to obtain the target type, location, and confidence level of the target to be detected.
[0048] To verify the effectiveness of this invention, in a specific application embodiment, the method described above was used to enhance and process a ground-penetrating radar (GPR) Bscan image, and the results were compared with those of a traditional YOLO image without CLAHE enhancement. The GPR device used was a GSSI SIR-4000 with a 1.5GHz antenna, an NVIDIA Jetson AGXXavier processing unit, and a detection speed of 23 FPS (640×480 resolution). The differences between the images before and after enhancement are shown below. Figure 2 As shown in Table 1, the differences in detection performance demonstrate that this invention can achieve accurate target positioning.
[0049] Table 1
[0050] A ground-penetrating radar-based target detection system includes: The data acquisition module is used to use ground penetrating radar to move along a straight line, detect the target to be detected during the operation, and acquire the echo data of the target to be detected at different detection points; The image processing module is used to normalize the echo data to obtain a normalized image, and convert the normalized image into a grayscale image to obtain a grayscale image. The splitting module is used to divide a grayscale image into several grayscale sub-images; The calculation module is used to obtain the frequency of occurrence of each gray level based on the gray level sub-image, and to obtain the cumulative distribution value based on the frequency of occurrence of the gray level; The adjustment module is used to crop the grayscale sub-image if the cumulative distribution value is greater than the preset threshold, adjust the grayscale level with the cumulative distribution value greater than the preset threshold and distribute it evenly to other grayscale levels, and calculate the cumulative distribution value of the adjusted grayscale sub-image until there is no cumulative distribution value greater than the preset threshold. The normalization module is used to normalize the cumulative distribution values of the adjusted grayscale sub-image to obtain a normalized value. The enhancement module is used to perform mapping based on the normalization value to enhance the contrast of the grayscale image and obtain the enhanced image. The image adjustment module is used to adjust the size and pixels of the enhanced image to obtain the adjusted image. The detection module is used to input the adjusted image into the pre-trained YOLO model to obtain the target type, location, and confidence level of the target to be detected, and to use the target type, location, and confidence level as the detection result.
[0051] The calculation module includes: The statistics unit is used to count the number of occurrences of each gray level based on the grayscale sub-image; The calculation unit is used to obtain the cumulative distribution value based on the number of occurrences of the gray level and the total number of pixels in the gray sub-image.
[0052] This application also discloses a terminal device, including a memory and a processor. The memory stores a computer program that can run on the processor. When the processor loads and executes the computer program, it employs a target detection method based on ground penetrating radar.
[0053] The terminal device can be a computer device such as a desktop computer, a laptop computer, or a cloud server. The terminal device includes, but is not limited to, a processor and a memory. For example, the terminal device may also include input / output devices, network access devices, and buses.
[0054] The processor can be a central processing unit (CPU). Of course, depending on the actual use, it can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), off-the-shelf programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor, etc., and this application does not limit it.
[0055] The memory can be an internal storage unit of the terminal device, such as a hard disk or RAM of the terminal device, or an external storage device of the terminal device, such as a plug-in hard disk, smart memory card (SMC), secure digital card (SD), or flash memory card (FC) equipped on the terminal device. Furthermore, the memory can be a combination of internal storage units and external storage devices of the terminal device. The memory is used to store computer programs and other programs and data required by the terminal device. The memory can also be used to temporarily store data that has been output or will be output. This application does not limit this.
[0056] In this terminal device, a ground-penetrating radar-based target detection method from the above embodiments is stored in the terminal device's memory and loaded and executed on the terminal device's processor for convenient use.
[0057] This application also discloses a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it employs a ground-penetrating radar-based target detection method as described in the above embodiments.
[0058] The computer program can be stored in a computer-readable medium. The computer program includes computer program code, which can be in the form of source code, object code, executable file, or certain middleware. The computer-readable medium includes any entity or device capable of carrying computer program code, recording media, USB flash drive, portable hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the computer-readable medium includes, but is not limited to, the above-mentioned components.
[0059] The above-described target detection method based on ground-penetrating radar is stored in the computer-readable storage medium and loaded and executed on the processor to facilitate the storage and application of the method.
[0060] Those skilled in the art should understand that the discussion of any of the above embodiments is merely exemplary and is not intended to imply that the scope of protection of this application is limited to these examples; within the framework of this application, the technical features of the above embodiments or different embodiments can also be combined, the steps can be implemented in any order, and there are many other variations of different aspects of one or more embodiments of this application as described above, which are not provided in detail for the sake of brevity.
[0061] One or more embodiments in this application are intended to cover all such substitutions, modifications, and variations that fall within the broad scope of this application. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of one or more embodiments in this application should be included within the protection scope of this application.
Claims
1. A target detection method based on ground-penetrating radar, characterized in that, include: The ground-penetrating radar travels in a straight line and detects the target during its operation, acquiring echo data of the target at different detection points. The echo data is normalized to obtain a normalized image, and the normalized image is converted into a grayscale image to obtain a grayscale image. The grayscale image is divided into several grayscale sub-images; Based on the grayscale subimage, the frequency of occurrence of each grayscale level is obtained, and the cumulative distribution value is calculated based on the frequency of occurrence of each grayscale level. If the cumulative distribution value is greater than a preset threshold, the grayscale sub-image is cropped, and the grayscale levels with cumulative distribution values greater than the preset threshold are adjusted and evenly distributed to other grayscale levels. The cumulative distribution value of the adjusted grayscale sub-image is calculated until there are no cumulative distribution values greater than the preset threshold. The cumulative distribution values of the adjusted grayscale sub-image are normalized to obtain the normalized value; Mapping is performed based on the normalization value to enhance the contrast of the grayscale image, resulting in an enhanced image. The enhanced image is then resized and pixel-adjusted to obtain the adjusted image; The adjusted image is input into a pre-trained YOLO model to obtain the target type, location, and confidence level of the target to be detected, and the target type, location, and confidence level are used as the detection result.
2. The target detection method based on ground-penetrating radar as described in claim 1, characterized in that, The step of normalizing the echo data to obtain normalized data, and converting the normalized data into a grayscale image, includes: The echo data is represented as a radar echo signal, as follows: in, For the first The coordinates of each detection point in the detection plane The total number of detection points, For time, and The ground-penetrating radar at the 1st The amplitude and phase of the echo received by each detection point; A two-dimensional image is drawn based on the radar echo signal; The two-dimensional image is normalized to obtain a normalized image; The normalized image is converted into a grayscale image to obtain a grayscale image.
3. The target detection method based on ground-penetrating radar as described in claim 1, characterized in that, The process of obtaining the frequency of occurrence of each gray level based on the gray-level sub-image and calculating the cumulative distribution value based on the frequency of occurrence of each gray level includes: Based on the grayscale sub-image, count the number of occurrences of each grayscale level; The cumulative distribution value is obtained based on the number of times the gray level appears and the total number of pixels in the gray sub-image.
4. The target detection method based on ground-penetrating radar as described in claim 1, characterized in that, The cumulative distribution values of the adjusted grayscale sub-image include: The number of remaining pixels after cropping is calculated based on the occurrence frequency and the cropping threshold. The number of occurrences after cropping is calculated based on the remaining number of pixels and the cropping threshold. The cumulative distribution value of the adjusted grayscale sub-image is calculated based on the number of occurrences after cropping.
5. The target detection method based on ground-penetrating radar as described in claim 1, characterized in that, The cumulative distribution value of the adjusted grayscale subimage is normalized to obtain a normalized value, which is expressed as: in, , This represents the cumulative distribution value of the adjusted grayscale subimage. It is the normalized value.
6. The target detection method based on ground-penetrating radar as described in claim 1, characterized in that, The step of inputting the adjusted image into a pre-trained YOLO model to obtain the target type, location, and confidence level of the target to be detected includes: Extract the features to be predicted from the adjusted image; The features to be predicted are input into a pre-trained YOLO model to obtain the target type, location, and confidence level of the target to be detected.
7. A target detection system based on ground-penetrating radar, characterized in that, include: The data acquisition module is used to use ground penetrating radar to move along a straight line, detect the target to be detected during the operation, and acquire the echo data of the target to be detected at different detection points; The image processing module is used to normalize the echo data to obtain a normalized image, and convert the normalized image into a grayscale image to obtain a grayscale image. A splitting module is used to divide the grayscale image into several grayscale sub-images; The calculation module is used to obtain the frequency of occurrence of each gray level based on the gray level sub-image, and to obtain the cumulative distribution value based on the frequency of occurrence of each gray level; The adjustment module is used to crop the grayscale sub-image if the cumulative distribution value is greater than a preset threshold, adjust the grayscale levels with cumulative distribution values greater than the preset threshold, and then evenly distribute them to other grayscale levels, and calculate the cumulative distribution value of the adjusted grayscale sub-image until there are no cumulative distribution values greater than the preset threshold. The normalization module is used to normalize the cumulative distribution values of the adjusted grayscale sub-image to obtain a normalized value. The enhancement module is used to perform mapping based on the normalization value to achieve contrast enhancement of the grayscale image, thereby obtaining an enhanced image; The image adjustment module is used to adjust the size and pixels of the enhanced image to obtain an adjusted image; The detection module is used to input the adjusted image into a pre-trained YOLO model to obtain the target type, location, and confidence level of the target to be detected, and to use the target type, location, and confidence level as the detection result.
8. The target detection system based on ground-penetrating radar as described in claim 1, characterized in that, The computing module includes: The statistics unit is used to count the number of occurrences of each gray level based on the grayscale sub-image; The calculation unit is used to obtain the cumulative distribution value based on the number of occurrences of the gray level and the total number of pixels in the gray sub-image.
9. A terminal device, comprising a memory and a processor, characterized in that, The memory stores a computer program that can run on a processor, and when the processor loads and executes the computer program, it employs the method described in any one of claims 1 to 7.
10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is loaded and executed by the processor, it employs the method described in any one of claims 1 to 7.