CN114782728A - Data visualization method for ground penetrating radar - Google Patents
Data visualization method for ground penetrating radar Download PDFInfo
- Publication number
- CN114782728A CN114782728A CN202210385145.2A CN202210385145A CN114782728A CN 114782728 A CN114782728 A CN 114782728A CN 202210385145 A CN202210385145 A CN 202210385145A CN 114782728 A CN114782728 A CN 114782728A
- Authority
- CN
- China
- Prior art keywords
- target
- penetrating radar
- ground penetrating
- data
- row
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S13/00—Systems using the reflection or reradiation of radio waves, e.g. radar systems; Analogous systems using reflection or reradiation of waves whose nature or wavelength is irrelevant or unspecified
- G01S13/88—Radar or analogous systems specially adapted for specific applications
- G01S13/885—Radar or analogous systems specially adapted for specific applications for ground probing
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
- G01S7/02—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00
- G01S7/41—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00 using analysis of echo signal for target characterisation; Target signature; Target cross-section
- G01S7/417—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00 using analysis of echo signal for target characterisation; Target signature; Target cross-section involving the use of neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Remote Sensing (AREA)
- Radar, Positioning & Navigation (AREA)
- General Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Biomedical Technology (AREA)
- Electromagnetism (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Computer Networks & Wireless Communication (AREA)
- Health & Medical Sciences (AREA)
- Evolutionary Biology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Geophysics And Detection Of Objects (AREA)
Abstract
The invention relates to a data visualization method of a ground penetrating radar, which comprises the following steps: 1) establishing an underground medium model, performing ground penetrating radar simulation to obtain random simulation data of a plurality of target parameters, and generating a corresponding B-scan image sample set; 2) performing ground penetrating radar target detection on the B-scan image to obtain vertex coordinates of a target signal, extracting a section near the vertex, establishing a corresponding data set according to material and size parameter data of a target corresponding to the section, and respectively corresponding to a pre-training neural network; 3) and inputting the newly extracted slice into a pre-trained neural network, estimating material and size parameters of the target, and printing corresponding parameter information on a corresponding vertex to complete data visualization. Compared with the prior art, the ground penetrating radar target detection method based on the line connection clustering algorithm is adopted, so that the ground penetrating radar target can be more accurately detected, particularly under the complex condition of target signal intersection, and is identified and visually displayed.
Description
Technical Field
The invention relates to the field of ground penetrating radar target detection and target identification, in particular to a data visualization method of a ground penetrating radar.
Background
Ground Penetrating Radar (GPR) is an important nondestructive evaluation device that detects underground areas using different electromagnetic properties of underground materials based on the propagation and scattering properties of high-frequency Electromagnetic (EM) waves, and nowadays, GPR, which is originally used in the military industry, has been deeply advanced into people's daily lives, such as engineering exploration and the like. Although the ground penetrating radar is very attractive as a nondestructive underground detection tool, the readability of ground penetrating radar data is unacceptable for people without abundant prior knowledge, if the timeliness of GPR use is far from being met by the manual interpretation of experts, the efficiency is relatively low, meanwhile, the manual interpretation usually depends on the professional knowledge and experience of a discriminator, and misjudgment and unnecessary errors are easy to occur. The data visualization requirements of ground penetrating radars are becoming more stringent, and due to the physical properties of the subsurface medium and the complexity of the electromagnetic wave propagation mechanism, the anomalies of the targets of the scanning area of the ground penetrating radar are often difficult to detect and identify automatically.
Although the existing method can achieve the purpose of target detection and identification on some simple targets, under the complex condition that hyperbola characteristics are crossed, because a common method cannot well separate crossed signals into independent target hyperbola signals, direct curve fitting often causes missed detection and false detection of targets, and the effect of the data visualization of the ground penetrating radar is greatly reduced.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provide a data visualization method for a ground penetrating radar.
The purpose of the invention can be realized by the following technical scheme:
a data visualization method for a ground penetrating radar comprises the following steps:
1) establishing an underground medium model, performing ground penetrating radar simulation to obtain random simulation data of a plurality of target parameters, and generating a corresponding B-scan image sample set;
2) performing ground penetrating radar target detection on the B-scan image to obtain vertex coordinates of a target signal, extracting a section near the vertex, establishing a corresponding data set according to material and size parameter data of a target corresponding to the section, and respectively corresponding to a pre-training neural network;
3) and inputting the newly extracted slice into a pre-trained neural network, estimating material and size parameters of the target, and printing corresponding parameter information on a corresponding vertex to complete data visualization.
In the step 1), ground penetrating radar simulation is carried out through a ground penetrating radar simulation method based on ray tracing to obtain simulation data with a plurality of random parameters, namely a B-scan image sample set is generated through gprMax software.
In the step 2), the neural network is specifically ResNet, the neural network takes the slices near the vertexes as input, and takes the material and size parameters of the corresponding target of the target as output for training.
In the step 2), the ground penetrating radar target detection is realized by adopting a ground penetrating radar target detection method based on a line connection clustering algorithm so as to obtain the vertex coordinates of the target signal.
The ground penetrating radar target detection method based on the row connection clustering algorithm specifically comprises the following steps:
21) preprocessing each original B-scan image in the B-scan image sample set, wherein the preprocessing comprises zero-time correction, ground clutter removal, amplitude enhancement and self-adaptive binarization processing to obtain two corresponding binarization images;
22) and extracting a target from each binary image, separating intersected hyperbolic target signals by adopting a line connection clustering algorithm, and fitting the separated hyperbolic signals to obtain the vertex position of a target hyperbolic curve to finish target detection.
In the step 21), zero time is acquired according to the frequency of the radar signal to carry out zero time correction;
the ground clutter removal is achieved by subtracting the average of each row, then:
where x (t) is the amplitude of a signal at time t, xj(t) is the amplitude of the jth B-scan signal at time t, n is the number of B-scan signals, and x1(t) is the amplitude of the B-scan signal at time t after ground clutter removal;
considering the energy attenuation in the propagation process, and adopting the linear time-varying gain to enhance the amplitude, there are:
where x2(t) is the amplitude of the amplitude enhanced signal at time t, tNThe total simulation duration is;
the binarization processing is carried out by adopting an adaptive threshold method, and the following steps are carried out:
threshold＝mean{Ie|Ie＞ρb×MaxIe}
wherein threshold is a threshold, mean { } is a function of the calculated mean, IeIs the intensity value of the edge pixel,
is the maximum value of the edge pixel intensity, pbIs a decimal between 0 and 1.
In the step 21), the B-scan image after amplitude enhancement is subjected to binarization processing twice to reduce information loss due to change of reflection polarity to the maximum extent, wherein the binarization processing is performed for the first time on the B-scan image after original amplitude enhancement and the other time on the B-scan image after reverse amplitude enhancement.
The step 22) specifically comprises the following steps:
221) using a cv2.findContours () function to separate potential blocks in the binary image, wherein the potential blocks are defined as bright blocks which are not connected with each other in the binary image;
222) separating intersection hyperbolas by adopting a row connection clustering algorithm for row segments in each potential block, wherein the row segments are defined as a continuous point set of each segment in a row of the potential block, and the method specifically comprises the following steps:
2221) judging whether the current line segment in the potential block meets the RCC condition, if so, performing step 2222), and if not, performing step 2223);
2222) connecting the row segment extension to the current target cluster;
2223) judging the segment of the row as a new stump cluster, transferring the judgment target to the next target cluster, and returning to the step 2221) to continuously judge whether the RCC condition is met;
2224) and if the line segment is not connected with all the existing target clusters, taking the line segment as a new target cluster.
223) And fitting according to the separated hyperbolas to obtain vertex position coordinates corresponding to the hyperbolas, and finishing target detection.
The RCC condition is provided based on the geometric characteristics of hyperbolas and is used for judging whether one row segment is a new stump caused by the intersection of other hyperbolas, and the RCC condition specifically comprises the following steps:
if the minimum column of a row segment is on the left side of the stump cluster to which it is connected, or the maximum column of the row segment is on the right side of the stump cluster to which it is connected, the RCC condition is considered to be satisfied, and if the row segment does not satisfy the RCC condition, it is considered as a new stump cluster.
In the step 2), in order to further eliminate bad samples caused by false detection of target detection before target identification, a positive and negative sample data set is established, and training of the neural network is carried out after negative samples are filtered out through a training positive and negative sample discrimination network.
Compared with the prior art, the invention has the following advantages:
the invention provides a full-flow processing scheme for interpreting and visualizing the data of the ground penetrating radar, which comprises preprocessing, target detection and target identification classification, realizes the visual output of the data of the ground penetrating radar to the target classification and parameters, adopts a row connection clustering algorithm in the target detection to improve the target detection capability, and compared with the existing column connection clustering algorithm, the invention directly captures hyperbolic vertices but not enough, thereby being capable of easily separating overlapped hyperbolic curves to accurately identify the target and visually displaying.
Drawings
Fig. 1 is a diagram illustrating the effect of the target detection in the embodiment of the present invention, wherein the pixels in the gray boxes are input to the network for positive and negative sample discrimination, and only the positive sample is recognized by the target.
FIG. 2 is an effect diagram of a data visualization method of the ground penetrating radar of the present invention.
Fig. 3 shows the relevant parameters of the materials involved in the model in the example.
FIG. 4 is a block diagram of a method flow of the present invention.
Detailed Description
The invention is described in detail below with reference to the figures and specific embodiments. The present embodiment is implemented on the premise of the technical solution of the present invention, and a detailed implementation manner and a specific operation process are given, but the scope of the present invention is not limited to the following embodiments.
The invention provides a data visualization method of a ground penetrating radar, which comprises the following steps:
1) establishing an underground medium model, performing ground penetrating radar simulation by using the conventional ground penetrating radar simulation method based on ray tracing to obtain a large amount of simulation data with random parameters, laying a foundation for establishing a sample set for subsequent target identification, namely generating a B-scan image sample set by using gprMax software;
2) and (3) realizing ground penetrating radar target detection by using a ground penetrating radar target detection method based on a line connection clustering algorithm to obtain the vertex coordinates of a target signal, and extracting a section near the vertex. And establishing a corresponding data set according to the material and the size of the target corresponding to the slice, and respectively pre-training a neural network (ResNet). In fact, in order to further eliminate bad samples caused by false detection in the target detection step before target identification, a positive and negative sample data set is also established for training a positive and negative sample discrimination network so as to filter out negative samples.
The method for detecting the ground penetrating radar target based on the row connection clustering algorithm is specifically realized by the following steps:
after the original B-SCAN image is converted into two binary images, sequentially extracting the target from each binary image.
Firstly, separating all hyperbolic regions from each other by using a Row-join clustering algorithm, and in order to better explain the Row-join clustering algorithm, defining the following three concepts including Row segments (Row segments), Potential blocks (Potential blocks) and RCC conditions, specifically:
line segment: each set of consecutive points in a row is called a row segment.
Potential blocks: there are some bright blocks in the binary image that are not connected to each other, and these bright blocks are called potential blocks.
RCC conditions: the RCC is used for judging whether one row segment is a new stump caused by other hyperbola intersections, and the condition is provided based on the geometrical characteristics of the hyperbolas, specifically, if the minimum column of the row segment is on the left side of the cluster connected with the row segment or the maximum column of the row segment is on the right side of the cluster connected with the row segment, the RCC is considered to be satisfied, and if the row segment does not satisfy the RCC, the row segment is considered as a new stump cluster.
The line segments in the potential block are then processed sequentially from top to bottom and from left to right. For each line segment, firstly judging whether the line segment is connected to the current cluster, if so, further judging whether RCC is met, and if so, expanding the line segment to the cluster; and if the RCC is not met, judging that the stump cluster is new, if the line segment is not connected with the clusters, performing the operation on the next cluster until the last cluster is processed, and if the line segment is not connected with all the existing clusters, regarding the line segment as a new cluster.
3) And (3) analyzing the section newly extracted by using the ground penetrating radar target detection method based on the row connection clustering algorithm by using a pre-trained neural network to estimate the material and size related parameters of the target, and printing related information on related vertexes of the simply processed ground penetrating radar data.
Therefore, the data visualization method of the ground penetrating radar is presented completely, so that the data visualization effect of the ground penetrating radar with high recognition rate can be obtained.
Examples
The data visualization method for the ground penetrating radar provided by the embodiment comprises the following steps:
1) establishing 850 models based on underground medium structures: the spatial dimensions of the model were 2 meters wide and 0.5 meters deep, the signal source was a Ricker pulse centered at 1.5GHz, the background of each model was filled with soil, three cylindrical objects were randomly placed in the soil, the underground objects had three material properties and three radius properties, their selection was also random, the materials included perfect conductor (PEC), polyvinyl chloride (PVC) and rock. The size categories included 1cm, 3cm and 5cm, with the parameters for each material as shown in FIG. 3.
2) The method comprises the steps of using a ground penetrating radar target detection method based on a row connection clustering algorithm to 850B-scan images generated by an underground medium structure model to generate 9429 hyperbolic vertex positive sample slices and 3419 hyperbolic vertex negative sample slices, expanding the positive and negative samples to 10000 through data enhancement, classifying the positive samples according to materials and radiuses, constructing a material data set and a radius data set, inputting the data into ResNet for training, wherein the target detection accuracy can reach 98.2%, and the average accuracy of material and radius identification can respectively reach 97.6% and 92.2%.
3) The pre-trained neural network is used for analyzing the slice newly extracted by the ground penetrating radar target detection method based on the line connection clustering algorithm, so that the material and size related parameters of the target can be estimated, and related information is printed on related vertexes of the ground penetrating radar data after simple processing, as shown in fig. 2, a plurality of obvious hyperbolic characteristics in the graph belong to three targets in the graph, 8 judgment sentences are generated, the second target has misjudgments on material identification and radius identification, but the misjudgments are easily discarded based on other judgment results of the same target: the number of times the same classification is judged and the classification probability thereof are important references.
The foregoing detailed description of the preferred embodiments of the invention has been presented. It should be understood that numerous modifications and variations could be devised by those skilled in the art in light of the present teachings without departing from the inventive concepts. Therefore, the technical solutions available to those skilled in the art through logic analysis, reasoning and limited experiments based on the prior art according to the concept of the present invention should be within the scope of protection defined by the claims.
Claims (10)
1. A data visualization method of a ground penetrating radar is characterized by comprising the following steps:
1) establishing an underground medium model, carrying out ground penetrating radar simulation to obtain simulation data with a plurality of random target parameters, and generating a corresponding B-scan image sample set;
2) performing ground penetrating radar target detection on the B-scan image to obtain vertex coordinates of a target signal, extracting slices near the vertex, establishing corresponding data sets according to material and size parameter data of the target corresponding to the slices, and respectively corresponding to a pre-training neural network;
3) and inputting the newly extracted slice into a pre-trained neural network, estimating material and size parameters of the target, and printing corresponding parameter information on a corresponding vertex to complete data visualization.
2. The method for visualizing data of a ground penetrating radar as recited in claim 1, wherein in said step 1), a ground penetrating radar simulation method based on ray tracing is used to perform ground penetrating radar simulation to obtain a plurality of simulation data with random parameters, i.e. a B-scan image sample set is generated by gprMax software.
3. The method as claimed in claim 1, wherein in step 2), the neural network is ResNet, and the neural network takes slices near the vertex as input and takes the material and size parameters of the target corresponding to the target as output for training.
4. The method for visualizing data of a ground penetrating radar as claimed in claim 1, wherein in the step 2), the ground penetrating radar target detection method based on the line connection clustering algorithm is adopted to realize the ground penetrating radar target detection so as to obtain the vertex coordinates of the target signal.
5. The method for visualizing data of a ground penetrating radar according to claim 4, wherein the method for detecting the target of the ground penetrating radar based on the line connection clustering algorithm specifically comprises the following steps:
21) preprocessing each original B-scan image in the B-scan image sample set, wherein the preprocessing comprises zero-time correction, ground clutter removal, amplitude enhancement and self-adaptive binarization processing to obtain two corresponding binarization images;
22) and extracting a target from each binary image, separating intersected hyperbolic target signals by adopting a line connection clustering algorithm, and fitting the separated hyperbolic signals to obtain the vertex position of a target hyperbolic curve to finish target detection.
6. The method for visualizing data of a ground penetrating radar as claimed in claim 5, wherein in said step 21), zero time correction is performed by acquiring zero time according to the frequency of radar signal;
the clutter removal is achieved by subtracting the average of each row, then:
where x (t) is the amplitude of a signal at time t, xj(t) is the amplitude of the jth B-scan signal at time t, n is the number of B-scan signals, and x1(t) is the amplitude of the B-scan signal at time t after ground clutter removal;
considering energy attenuation in the propagation process, and adopting linear time-varying gain to carry out amplitude enhancement, there are:
where x2(t) is the amplitude of the amplitude enhanced signal at time t, tNThe total simulation duration is;
the binarization processing is carried out by adopting an adaptive threshold method, and the following steps are carried out:
wherein threshold is a threshold, mean { } is a function of the calculated mean, IeIs the intensity value of the edge pixel,
is the maximum value of the edge pixel intensity, pbA decimal fraction between 0 and 1.
7. The method as claimed in claim 6, wherein in step 21), the amplitude-enhanced B-scan image is binarized twice to minimize information loss due to the change of the reflection polarity, wherein the first time is applied to the original amplitude-enhanced B-scan image, and the second time is applied to the inverted amplitude-enhanced B-scan image.
8. The method for visualizing data of a ground penetrating radar as recited in claim 5, wherein said step 22) comprises the following steps:
221) using a cv2.findContours () function to separate potential blocks in the binary image, wherein the potential blocks are defined as bright blocks which are not connected with each other in the binary image;
222) separating the hyperbolae of intersection by adopting a row connection clustering algorithm for the row segment in each potential block, wherein the row segment is defined as a continuous point set of each segment in one row of the potential block, and the method specifically comprises the following steps:
2221) judging whether the current line segment in the potential block meets the RCC condition, if so, performing step 2222), and if not, performing step 2223);
2222) connecting the row segment extension to the current target cluster;
2223) judging the segment of the row as a new stump cluster, transferring the judgment target to the next target cluster, and returning to 2221) to continuously judge whether the RCC condition is met;
2224) if the line segment is not connected with all the existing target clusters, the line segment is taken as a new target cluster.
223) And fitting according to the separated hyperbolas to obtain vertex position coordinates corresponding to the hyperbolas, and completing target detection.
9. The method for visualizing data of a ground penetrating radar according to claim 8, wherein the RCC condition is provided based on geometric features of hyperbolas, and is used for determining whether a segment of a row is a new stump caused by intersection of other hyperbolas, specifically:
if the minimum column of a row segment is on the left side of the stump cluster to which it is connected, or the maximum column of the row segment is on the right side of the stump cluster to which it is connected, the RCC condition is considered to be satisfied, and if the row segment does not satisfy the RCC condition, it is considered as a new stump cluster.
10. The method as claimed in claim 1, wherein in step 2), in order to further eliminate bad samples caused by false detection of target detection before target recognition, a positive and negative sample data set is established, and training of the neural network is performed after negative samples are filtered out by training a positive and negative sample discrimination network.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210385145.2A CN114782728A (en) | 2022-04-13 | 2022-04-13 | Data visualization method for ground penetrating radar |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210385145.2A CN114782728A (en) | 2022-04-13 | 2022-04-13 | Data visualization method for ground penetrating radar |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114782728A true CN114782728A (en) | 2022-07-22 |
Family
ID=82428769
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210385145.2A Pending CN114782728A (en) | 2022-04-13 | 2022-04-13 | Data visualization method for ground penetrating radar |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114782728A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116297544A (en) * | 2023-03-16 | 2023-06-23 | 南京京烁雷达科技有限公司 | Method and device for extracting target object of coal rock identification ground penetrating radar |
| CN116539643A (en) * | 2023-03-16 | 2023-08-04 | 南京京烁雷达科技有限公司 | Method and system for identifying coal rock data by using radar |
-
2022
- 2022-04-13 CN CN202210385145.2A patent/CN114782728A/en active Pending
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116297544A (en) * | 2023-03-16 | 2023-06-23 | 南京京烁雷达科技有限公司 | Method and device for extracting target object of coal rock identification ground penetrating radar |
| CN116539643A (en) * | 2023-03-16 | 2023-08-04 | 南京京烁雷达科技有限公司 | Method and system for identifying coal rock data by using radar |
| CN116539643B (en) * | 2023-03-16 | 2023-11-17 | 南京京烁雷达科技有限公司 | Method and system for identifying coal rock data by using radar |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2021243743A1 (en) | Deep convolutional neural network-based submerged oil sonar detection image recognition method | |
| Merazi-Meksen et al. | Mathematical morphology for TOFD image analysis and automatic crack detection | |
| US20230108634A1 (en) | Ground penetrating radar and deep learning-based underground pipeline detection method and system | |
| Lee et al. | Automatic target detection in GPR images using Histogram of Oriented Gradients (HOG) | |
| CN114782728A (en) | Data visualization method for ground penetrating radar | |
| CN107389701A (en) | A kind of PCB visual defects automatic checkout system and method based on image | |
| CN107392141B (en) | Airport extraction method based on significance detection and LSD (least squares distortion) line detection | |
| CN105447851A (en) | Glass panel sound hole defect detection method and system | |
| CN102928435A (en) | Aircraft skin damage identification method and device based on image and ultrasound information fusion | |
| CN110766689A (en) | Method and device for detecting article image defects based on convolutional neural network | |
| Noreen et al. | Using pattern recognition with HOG to automatically detect reflection hyperbolas in ground penetrating radar data | |
| Gui et al. | Automated defect detection and visualization for the robotic airport runway inspection | |
| CN113359101B (en) | Underground target detection method, system and computer storage medium | |
| CN109003275A (en) | The dividing method of weld defect image | |
| CN111091071B (en) | Underground target detection method and system based on ground penetrating radar hyperbolic wave fitting | |
| CN112329587A (en) | Beverage bottle classification method and device and electronic equipment | |
| Sinha et al. | Morphological segmentation and classification of underground pipe images | |
| Mamatov et al. | Algorithms for contour detection in agricultural images | |
| CN114942475A (en) | Ground penetrating radar target detection method based on row connection clustering algorithm | |
| Shihab et al. | Image processing and neural network techniques for automatic detection and interpretation of ground penetrating radar data | |
| Liu et al. | Advances in automatic identification of road subsurface distress using ground penetrating radar: State of the art and future trends | |
| CN115409768A (en) | Water leakage disease deep learning detection method based on characteristic input enhancement | |
| CN113534146A (en) | Radar video image target automatic detection method and system | |
| 迟大钊 et al. | Defect detection method based on 2D entropy image segmentation | |
| Zhou et al. | Research on Traditional Image Segmentation Method Based on Oil Drilling Pipe Defects |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |