Device and method for discriminating buried object

Abstract

This device for discriminating a buried object is configured: to receive input of underground exploration data provided from the outside, and to estimate the permittivity of an underground medium on the basis of the input underground exploration data; to execute aperture synthesis processing on the underground exploration data on the basis of the estimated permittivity; to conduct discrimination processing for discriminating a buried object on the basis of the aperture synthesis exploration data obtained as a result of the aperture synthesis processing conducted on the underground exploration data and on the basis of the underground exploration data; and to output, as buried object information, information related to the buried object obtained from the discrimination processing.

Description

Buried object discrimination device and method

[0001] The present invention relates to a buried object discrimination device and method, and is suitable for application to a buried object discrimination device that analyzes underground exploration data acquired by a ground penetrating radar device to discriminate underground buried objects (hereinafter simply referred to as buried objects).

[0002] Conventionally, when erecting electric poles or removing electric poles, surveys of buried objects in the construction area are carried out. Test excavation is costly and can damage the buried objects during excavation. Therefore, underground radar is used to explore buried objects as a low-cost, non-destructive survey method.

[0003] Electromagnetic waves emitted from a ground-penetrating radar device usually propagate underground at a certain radiation angle. Therefore, the reflected waves received by the ground-penetrating radar device include not only those from buried objects located directly below the device, but also those from buried objects located in the surrounding area. Therefore, the data acquired from multiple points on the ground surface during the detection of buried objects contains distorted images reflected from buried objects.

[0004] In manual buried object surveys, underground exploration data acquired by ground-penetrating radar equipment is converted into images, and the characteristic images representing signals from the buried objects are then interpreted by the human eye to investigate the buried objects.

[0005] Non-Patent Document 2 discloses a technique for estimating buried depth and electromagnetic wave velocity in a medium from underground exploration data. This technique estimates values ​​so that the theoretical arrival time fits to the clear reflection image in the underground exploration data by changing the velocity of the electromagnetic wave in the medium and the buried object depth. Furthermore, Non-Patent Document 3 discloses a technique for correcting errors due to the angle of the lateral line crossing a buried pipe.

[0006] Non-Patent Document 1 discloses a synthetic aperture processing technique that uses data acquired at multiple points on the ground surface. Synthetic aperture processing is a technique that synthesizes reflected signals from the same buried object at the reflection point based on parameters such as the relative positions of the transmitting and receiving antennas of the underground radar device, the sampling interval during the survey, and the relative dielectric constant of the ground. This makes it possible to correct the characteristic image distortion mentioned above and generate synthetic aperture survey data with an improved signal-to-noise ratio. Distortion correction is useful for improving the accuracy of determining the location and size of buried objects.

[0007] Non-Patent Document 4 discloses a technique for applying the above-mentioned synthetic aperture processing technology to determine the dielectric constant of the ground. Images corrected by synthetic aperture processing converge to the reflection point, but if errors are included in the parameters used in the calculation, such as the dielectric constant, distortion of the image remains. The results of synthetic aperture processing are evaluated using a numerical index based on the degree of convergence, and the dielectric constant at which the evaluation result is maximized is searched for, thereby appropriately obtaining the average dielectric constant around the image.

[0008] Patent Literature 1 discloses a method for estimating the depth of multiple buried objects in non-uniform soil. In non-uniform soil, the dielectric constant varies depending on the location due to the soil composition. This technology performs a migration process, a type of aperture synthesis, on underground exploration data while varying the dielectric constant within a certain range. The coordinate value and dielectric constant at which the migration result is maximized are considered to be the average dielectric constant at those coordinates and extracted. The location of the buried objects is then detected using a pre-trained buried object detector. The buried object detector learns using a machine learning method used for image recognition. The migration results are referenced at the coordinates surrounding the detected buried object to determine the corresponding dielectric constant. The depth of the buried object is then estimated from the coordinates and dielectric constant of the buried object.

[0009] Furthermore, Patent Literature 2 discloses a technology for automatically identifying buried objects using a buried object learning model based on underground exploration images acquired by an underground radar device. The buried object learning model is an AI (Artificial Intelligence) model configured by a neural network that has learned underground exploration data of the detection target. This technology is characterized in particular by generating an underground exploration image from externally provided underground exploration data and synthetic aperture exploration data that has been subjected to aperture synthesis processing, and identifying buried objects based on the generated underground exploration image.

[0010] International Publication No. 2022 / 264342 Japanese Patent Application Laid-Open No. 2022-1203327

[0011] Matsuo Sekine, "Radar Signal Processing Technology," First Edition, Institute of Electronics, Information and Communication Engineers, September 20, 1991. Motoyuki Sato, "Geophysical Exploration Seminar 'Ground Penetrating Radar'," [online], July 2001, Center for Northeast Asian Studies, Tohoku University, [Retrieved June 13, 2023], Internet <URL: http: / / cobalt.cneas.tohoku.ac.jp / users / sato / GroundPenetratingRadarSubsurfaceMeasurement(2001)revised.pdf>. WWL Lai, JFC Sham, and F. Xie, "Correction of GPR wavevelocity with distorted hyperbolic reflection in underground utility's GPR survey," 2016 16th International Conference on Ground Penetrating Radar (GPR), Hong Kong, China, 2016, pp. 1-4. X. Wei and Y. Zhang, "Autofocusing Techniques for GPR Data from RC Bridge Decks," in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, vol. 7, no. 12, pp. 4860-4868, Dec. 2014

[0012] According to the invention disclosed in Patent Document 2, highly accurate buried object identification can be achieved by inputting synthetic aperture processing data in addition to underground exploration data, generating an integrated image, and then inputting it into an identifier. This is because high-quality aperture processing data based on appropriate parameters is expected to correct distortion in the reflected image of the underground exploration data and improve the signal-to-noise ratio. However, because the synthetic aperture processing data is generated outside the system and is not connected to the buried object identifier, there is a possibility that external synthetic aperture processing data that does not meet the input specifications of the buried object identifier may be generated, which means that it cannot be input to the buried object identifier together with the underground exploration data.

[0013] In addition, as mentioned above, aperture synthesis processing requires device-specific parameters, such as the relative positional relationship between the transmitting antenna and the receiving antenna of the underground radar device, as well as the dielectric constant of the soil. Therefore, aperture synthesis processing performed outside the system is necessarily performed independently for each underground radar device and each search area. Therefore, when aperture synthesis processing is performed in an external system, it is not easy to use the aperture synthesis data in the same system.

[0014] In this regard, if the format of the input data to the identifier that identifies buried objects can be guaranteed within the device, aperture synthesis processing data can be reliably input to the identifier, the identification performance of buried objects can be improved, and buried objects can thus be identified with high reliability.

[0015] In the invention disclosed in Patent Document 1, underground exploration data is subjected to aperture synthesis processing in advance using multiple relative dielectric constants, and the depth of buried objects is estimated by referencing the results with the position of buried objects detected by a separately prepared buried object detector.

[0016] However, after the buried object detector has finished detecting the buried object locations, the relative permittivity candidates for each buried object location are identified. Therefore, there is a problem in that it is not possible to input synthetic aperture survey data that has been subjected to synthetic aperture processing using an appropriate relative permittivity into the buried object detector. In addition, it is necessary to perform synthetic aperture processing multiple times for the number of predetermined relative permittivities, which poses a problem in balancing the calculation cost with the estimation resolution of the relative permittivity (and the depth of the buried object converted from it).

[0017] The present invention has been made in consideration of the above points, and proposes a buried object discrimination device and method that can discriminate buried objects with high reliability.

[0018] In order to solve this problem, in the present invention, a buried object discrimination device that discriminates buried objects underground based on underground exploration data obtained by transmitting and receiving electromagnetic waves underground is provided with a dielectric constant estimation unit that estimates the dielectric constant of the underground medium based on the underground exploration data provided from outside and outputs an estimated dielectric constant, an aperture synthesis processing unit that performs aperture synthesis processing on the underground exploration data using the estimated dielectric constant and outputs synthetic aperture exploration data, and a buried object discrimination unit that performs discrimination processing to identify the buried object based on the aperture synthesis exploration data and the underground exploration data and outputs information about the buried object as buried object information.

[0019] In addition, the present invention provides a buried object discrimination method executed by a buried object discrimination device that discriminates buried objects underground based on underground exploration data obtained by transmitting and receiving electromagnetic waves underground, the method comprising: a first step of inputting the underground exploration data provided from outside; a second step of estimating the relative dielectric constant of the underground medium based on the input underground exploration data; a third step of performing aperture synthesis processing on the underground exploration data based on the estimated relative dielectric constant; a fourth step of performing an identification process to identify the buried object based on the aperture synthesis exploration data obtained by performing the aperture synthesis processing on the underground exploration data and the underground exploration data; and a fifth step of outputting information related to the buried object obtained by the identification process as buried object information.

[0020] According to the buried object discrimination device and method of the present invention, a processing section that estimates the relative dielectric constant of the underground medium based on underground exploration data and a processing section that performs aperture synthesis processing on the same underground exploration data using the estimated relative dielectric constant of the underground medium are combined with a buried object identification processing section, so that the aperture synthesis exploration data can be reliably input to the buried object identification processing section, making it possible to increase the buried object recognition rate.Furthermore, the detection rate of the buried object identification section that uses the results can be improved and the false recognition rate can be reduced.

[0021] Furthermore, in the present invention, underground exploration data containing, as antenna structure information, information about the antenna structure of the underground radar device that generated the underground exploration data is input, and at least one of the dielectric constant estimation unit and the aperture synthesis processing unit estimates the dielectric constant based on a radar reflection formula corresponding to the antenna structure based on the antenna structure information contained in the underground exploration data, among radar reflection formulas previously defined for each antenna structure, or performs aperture synthesis processing on the underground exploration data, and inputs the result to the buried object identification unit. Here, the radar reflection formula is defined as a model that expresses the time it takes for an electromagnetic wave emitted from the underground radar device to be reflected by a buried object and received by a receiving antenna.

[0022] According to the buried object discrimination device of the present invention, the accuracy of the dielectric constant estimation by the dielectric constant estimation unit and the aperture synthesis processing by the aperture synthesis processing unit can be maintained constant even for underground radar devices with different antenna structures, and the estimated or generated dielectric constant and synthetic aperture exploration data can be maintained to the maximum extent possible as a buried object discrimination device.

[0023] Furthermore, in the present invention, the buried object discrimination device is provided with a depth conversion unit. The buried object information output from the buried object identification unit includes the position of the buried object on the time axis of the underground exploration data. The depth conversion unit converts the position on the time axis into a depth representing the distance of the buried object from the ground surface based on the estimated relative dielectric constant estimated by the relative dielectric constant estimation unit.

[0024] In addition, after the buried object identification process, the relative dielectric constant estimation unit re-estimates the relative dielectric constant based on the buried object information, and the depth conversion unit converts the depth of the buried object into a depth based on the re-estimated relative dielectric constant.

[0025] According to the buried object discrimination device of the present invention, the relative dielectric constant determined in advance for use in aperture synthesis processing can be reused within the device, and depth can be added to the buried object information output from the buried object discrimination device, allowing the user to intuitively confirm the depth position of the buried object.

[0026] Furthermore, in the present invention, the aperture synthesis processing unit re-executes the aperture synthesis processing based on an aperture synthesis processing re-execution flag input from outside, using the relative dielectric constant re-estimated by the relative dielectric constant estimation unit, the buried object identification unit re-executes the buried object identification processing based on the aperture synthesis exploration data obtained by the aperture synthesis processing re-execution by the aperture synthesis processing unit, and the depth conversion unit converts the depth of the buried object to the depth of the buried object based on the buried object information obtained by the buried object identification processing re-execution and the relative dielectric constant re-estimated by the relative dielectric constant estimation unit.

[0027] The buried object discrimination device of the present invention can regenerate synthetic aperture detection data to be input to the buried object discrimination unit based on the high-level buried object discrimination results of the buried object discrimination unit. This improves the convergence and S / N of the synthetic aperture detection data input to the buried object discrimination unit, thereby improving the buried object discrimination accuracy of the buried object discrimination unit.

[0028] Furthermore, in the present invention, a selection unit is provided that has the function of determining whether or not to re-execute the aperture synthesis processing in the aperture processing unit, the buried object identification processing in the buried object identification unit, or the depth conversion processing in the depth conversion unit.

[0029] Since the identification results of buried objects using synthetic aperture exploration data updated by regeneration as input may be updated, this identification result may be returned to the relative dielectric constant estimation unit, and the relative dielectric constant may be re-estimated in the relative dielectric constant estimation unit and the aperture synthesis processing may be performed in the synthetic aperture processing unit multiple times.

[0030] By doing so, the convergence and S / N of the synthetic aperture survey data input to the buried object identification unit are improved, and the identification accuracy of the buried object identification unit is improved. In this process, the selection unit that determines whether or not the synthetic aperture processing, buried object identification processing, or depth conversion processing needs to be performed again can automatically improve the reliability of buried object identification in the buried object discrimination device of the present invention, making it possible to perform the process with even greater reliability.

[0031] According to the present invention, it is possible to realize a buried object discrimination device and method that can discriminate buried objects with high reliability.

[0032] 1 is a block diagram showing the configuration of a buried object discrimination device according to claim 1. FIG. 2 is a block diagram showing the configuration of a buried object management system according to a first embodiment. (A) is a diagram explaining underground exploration data, and (B) is a diagram explaining synthetic aperture exploration data. (B) is a diagram explaining three-dimensional underground exploration data. (C) is a diagram explaining a buried object identification process using a CNN model. (A) shows the relationship between the CNN model and input data, (B) shows a combination of underground exploration data and synthetic aperture exploration data, and (C) shows a method of using underground exploration data and synthetic aperture data for identification. FIG. 3 is a flowchart of a buried object discrimination method according to claim 14. (A) is an example of buried object information for two-dimensional underground exploration data, and (B) is an example of buried object information for three-dimensional underground exploration data. (A) is an example of buried object information with depth for two-dimensional underground exploration data, and (B) is an example of buried object information with depth for three-dimensional underground exploration data. FIG. 4 is a block diagram of a buried object management system according to claim 13. 10 is a diagram showing an example of the screen configuration of an underground exploration result screen according to the second embodiment. FIG. 11 is a block diagram showing the configuration of a buried object management system according to the third embodiment. FIG. 12 is a block diagram showing the configuration of a buried object discrimination device according to claim 2. FIG. 13 is a diagram showing an example of the screen configuration of an underground exploration result screen according to the third embodiment. FIG. 14 is a flowchart showing the processing procedure for buried object discrimination processing according to the fourth embodiment. FIG. 15 is a block diagram showing the configuration of a buried object discrimination device according to claim 3. FIG. 16 is a block diagram showing the configuration of a buried object management system according to the fourth embodiment. FIG. 17 is a diagram showing an example of the configuration of a buried object discrimination device according to claim 4, and FIG. 18 is a block diagram showing the configuration of a selection unit according to claim 4. FIG. 18 is a block diagram showing the configuration of a buried object management system according to a fifth embodiment. FIG. 19 is a table showing an example of how a re-execution flag is determined by the selection unit according to claim 4 in the sixth embodiment. FIG. 19 is a block diagram showing the configuration of a selection unit according to claim 5. FIG. 19 is a block diagram showing the configuration of a buried object discrimination device according to claim 6, and FIG. 19 is a block diagram showing the configuration of a buried object learning model generation unit according to claim 6. FIG. 19 is a block diagram showing the configuration of a buried object management system according to a seventh embodiment. FIG. 19 is a block diagram showing the configuration of a buried object learning model generation unit according to claim 7.10 is a block diagram showing the configuration of a buried object discrimination device according to claim 8. FIG. 11 is a block diagram showing the configuration of a buried object management system according to the eighth embodiment. FIG. 12 is a block diagram showing the configuration of a buried object discrimination device according to claim 9. FIG. 13 is a block diagram showing the configuration of a buried object management system according to the ninth embodiment. FIG. 14 is a block diagram showing the configuration of a buried object discrimination device according to claim 10. FIG. 15 is a block diagram showing the configuration of a buried object management system according to the tenth embodiment. FIG. 16 is a diagram showing the configuration of a buried object management system according to the tenth embodiment. (A-1) to (B-4) are diagrams used to explain differences in radar reflection types due to differences in antenna structure. FIG. 17 is a block diagram showing the configuration of a buried object discrimination device according to claim 11. FIG. 18 is a block diagram showing the configuration of a buried object management system according to the eleventh embodiment. FIG. 19 is a flowchart showing the processing steps of buried object discrimination processing by a buried object discrimination device according to the eleventh embodiment. FIG. 19 is a flowchart showing the processing steps of buried object discrimination processing by a buried object discrimination device according to the eleventh embodiment. FIG. 19 is a block diagram showing the configuration of a buried object learning model generation unit according to claim 12. FIG. 19 is a block diagram showing the configuration of a buried object management system according to the twelfth embodiment.

[0033] An embodiment of the present invention will be described in detail below with reference to the drawings.

[0034] (1) First Embodiment (1-1) Configuration of a Buried Object Management System According to this Embodiment In Fig. 1B, the buried object management system according to this embodiment is generally designated by 1. This buried object management system 1 is configured to include a buried object discrimination device 2 and an information display device 3.

[0035] The buried object discrimination device 2 is composed of a general-purpose computer device equipped with information processing resources such as a CPU (Central Processing Unit) and memory. The buried object discrimination device 2 receives underground exploration data D1 acquired by an underground radar device (not shown), discriminates the type, shape, position, size, etc. of the buried object based on the underground exploration data D1, and outputs the discrimination results to the information display device 3 as buried object information D2.

[0036] The information display device 3 is configured with a display device equipped with, for example, a liquid crystal display or an organic EL (Electro-Luminescence) display, etc. The information display device 3 generates a predetermined inspection result screen on which the buried object discrimination results are displayed based on the buried object information D2 provided by the buried object discrimination device 2, and displays the generated inspection result screen.

[0037] (1-2) Configuration of the buried object discrimination device according to this embodiment As shown in FIG. 1A, the buried object discrimination device 2 according to this embodiment is configured to include a relative dielectric constant estimation unit 11, an aperture synthesis processing unit 12, and a buried object identification unit 13.

[0038] The relative dielectric constant estimation unit 11, the aperture synthesis processing unit 12, and the buried object identification unit 13 are realized by the above-mentioned CPU provided in the buried object discrimination device 2 executing a program stored in the above-mentioned memory provided in the buried object discrimination device 2.

[0039] The underground exploration data D1 is data acquired by transmitting and receiving electromagnetic waves underground using an underground radar device, and includes intensity information and phase information of the received signal for the time interval between transmission and reception of the electromagnetic waves.

[0040] Fig. 2A shows an example of such underground exploration data D1. In Fig. 2A, the vertical axis represents the time (t) from when the underground radar device emits electromagnetic waves into the ground until the underground radar device receives the waves reflected from the ground, and the horizontal axis represents the distance (x) in the exploration direction (the direction of travel of the underground radar device). Furthermore, brightness represents signal strength. The signal strength may be a positive or negative value, or even an imaginary number.

[0041] In Figure 2 (A), the underground exploration data D1 is shown as a two-dimensional image with a vertical axis and a horizontal axis, but the underground exploration data D1 obtained, for example, by an array-type underground radar device equipped with multiple transmitting antennas and receiving antennas becomes three-dimensional data that includes multiple two-dimensional data shown in Figure 2 (A).

[0042] As shown in FIG. 3, for example, the three-dimensional data has axes of time (t) and the direction of travel of the survey (x), as well as an axis of the vertical direction (Y) of the direction of travel of the survey.

[0043] The dielectric constant estimation unit 11 in FIGS. 1A and 1B is a functional unit that has a function of estimating the dielectric constant of an underground medium based on underground exploration data D1 input from outside, and outputting the estimated dielectric constant E1 to the aperture synthesis processing unit 12.

[0044] The relative dielectric constant can be estimated by applying a fitting method, as disclosed in Non-Patent Document 2, in which clear points (points with high signal strength) in an image based on underground exploration data D1 are extracted as reflection points of the electromagnetic waves emitted from the underground radar device at buried objects, and parameters are changed based on the coordinates of the extracted clear points (reflection points) and the underground exploration data D1 so that the theoretical arrival time fits the underground exploration data D1.

[0045] As a fitting method for changing parameters so as to fit the theoretical arrival time to the underground exploration data D1, for example, the fitting method disclosed in Non-Patent Document 3 can be applied.

[0046] If no clear reflection point is found in the underground exploration data D1, the dielectric constant estimation unit 11 may output a predetermined dielectric constant. In this case, the "predetermined dielectric constant" is usually a range of dielectric constants that can actually be assumed depending on the soil and moisture that make up most of the underground, which is given to the dielectric constant estimation unit 11 in advance. Since the dielectric constant varies depending on the location underground, multiple dielectric constants may be estimated in the underground exploration data D1, and in such a case, the dielectric constant estimation unit 11 may output multiple estimated dielectric constants E1.

[0047] The synthetic aperture processing unit 12 is a functional unit having a function of performing a predetermined synthetic aperture processing on the input underground exploration data D1 using the estimated dielectric constant E1 of the underground medium provided by the dielectric constant estimation unit 11. The synthetic aperture processing unit 12 outputs synthetic aperture exploration data D3, as shown in Fig. 2(B), obtained by the synthetic aperture processing to the buried object identification unit 13. When there are multiple estimated dielectric constants E1 provided by the dielectric constant estimation unit 11, statistical values ​​such as the average, median, maximum, or minimum value of the estimated dielectric constants E1 may be calculated and then used.

[0048] As an example of such aperture synthesis processing, it is possible to apply the technology described on pages 262 to 267 of Non-Patent Document 1. In addition, the predetermined migration processing disclosed in Patent Document 1 is also a type of aperture synthesis processing and can be applied.

[0049] Because electromagnetic waves emitted into the ground spread at a fixed radiation angle, the receiving antenna also receives reflected waves from buried objects not located directly below the ground. For this reason, the underground exploration data D1 usually contains distortion. Because this distortion can be corrected by synthetic aperture processing, it is useful for determining the horizontal and depth positions of buried objects. Inputting the synthetic aperture exploration data D3 into the buried object identification unit 13 also improves the accuracy of buried object identification by the buried object identification unit 13.

[0050] Furthermore, when the aperture synthesis processing unit 12 is given the estimated relative dielectric constant E1 for each location estimated by the relative dielectric constant estimation unit 11, the aperture synthesis processing unit 12 may perform partial aperture synthesis processing on a predetermined data range including each location and output integrated data. In this case, if the data ranges overlap, a representative value such as an average value may be calculated for the overlapping ranges and integrated.

[0051] The buried object identification unit 13 is a functional unit that has the function of extracting reflected images originating from buried objects based on the input underground exploration data D1 and the synthetic aperture exploration data D3 provided by the synthetic aperture processing unit 12, and deriving buried object information D2 including information on the type, shape, position, size, etc. of the buried object. The buried object information D2 derived by the buried object identification unit 13 is output from the buried object discrimination device.

[0052] The buried object identification unit 13 may derive buried object information using a buried object learning model, which is an AI-based learning model for buried objects, or may derive buried object information using a rule base or multidimensional clustering. Furthermore, the buried object identification unit 13 may derive buried object information D2 identified by multiple buried object learning models.

[0053] As an AI-based buried object learning model, for example, a convolutional neural network (CNN) model can be used. An example of the configuration of a buried object identification unit using a CNN model is shown in FIG. 4.

[0054] The CNN model is shown in FIG. 4A and may include a convolutional layer and a pooling layer. In the convolutional layer, input data is convolved using a specified filter, and the resulting values ​​(feature map data) are output to the pooling layer. Activation using an activation function may be performed before the feature map data is output from the convolutional layer to the pooling layer. Examples of the activation function that may be used include a sigmoid function, a softmax function, and a ReLU (Rectified Linear Unit).

[0055] The pooling layer is provided after the convolution layer and reduces the amount of data input from the convolution layer using a specified filter before outputting the feature map data. The maximum or average value can be used, for example, to reduce the amount of data. The fully connected layer is used as the output layer in the CNN model and combines and outputs the feature map data. A flatten function can be used to combine the feature map data.

[0056] In order to input the underground exploration data D1 and the synthetic aperture exploration data D3 into the CNN model, the underground exploration data D1 and the synthetic aperture exploration data D3 can be synthesized in advance by a synthesis processing unit to generate synthesized data D4, as shown in Fig. 4(B) . In this case, methods for synthesizing the underground exploration data D1 and the synthetic aperture exploration data D3 can include, for example, taking the average value, linear sum, maximum value, or minimum value of the data values ​​at each coordinate.

[0057] 4(C), the buried object identifier 13 derives buried object information D21, D22 based on the underground exploration data D1 and the synthetic aperture exploration data D3 using identifiers (CNN models A, B) prepared corresponding to the underground exploration data D1 and the synthetic aperture exploration data D3, respectively, and outputs the buried object information D21, D22 derived based on the underground exploration data D1 and the synthetic aperture exploration data D3, respectively, to a result integrator. The result integrator may select either or both of the buried object information D21, D22, or information common to the buried object information D21, D22, and output it as buried object information D2.

[0058] By preparing classifiers corresponding to the underground exploration data D1 and the synthetic aperture exploration data D3, respectively, it is possible to perform classification suited to the feature quantities possessed by the underground exploration data D1 and the synthetic aperture exploration data D3, and to obtain buried object information D2 corresponding to these features. Then, by mutually complementing each other based on this buried object information D2, it is possible to improve the recall rate and precision rate of the output of the buried object identification unit 13.

[0059] The buried object information D2 derived by the buried object identification unit 13 is output to the information display device 3, and a predetermined exploration result screen containing information such as the type, shape, location, and size of the buried object contained in the buried object information D2 is displayed on the information display device 3.

[0060] An example of buried object information D2 displayed on the information display device 3 will be described with reference to Figure 6. Figure 6(A) is an example of buried object information for two-dimensional underground exploration data. The position in the buried object information D2 is represented by the position relative to the direction of exploration and the time it takes for the underground radar device to receive the reflected wave from the ground. A point object may be represented by a single point, and a line object may be represented by a line segment connecting the start point and end point. Figure 6(B) is an example of buried object information for three-dimensional underground exploration data. In this case, the buried object information D2 also has a component perpendicular to the direction of exploration.

[0061] FIG. 5 shows a buried object discrimination method in the buried object discriminating device 2 having such a configuration.

[0062] First, underground exploration data D1 is input (S1).

[0063] Next, the dielectric constant estimation unit 11 estimates the estimated dielectric constant E1 of the underground medium based on the input underground exploration data D1, and outputs the estimated dielectric constant E1 to the aperture synthesis processing unit 12 (S2). Furthermore, the aperture synthesis processing unit 12 performs aperture synthesis processing on the input underground exploration data D1 using the estimated dielectric constant E1 provided by the dielectric constant estimation unit 11, and outputs the above-mentioned synthetic aperture exploration data D3 thus obtained to the buried object identification unit 13 (S3).

[0064] The buried object identification unit 13 identifies buried objects based on the input underground exploration data D1 and the aperture synthesis exploration data D3 provided by the aperture synthesis processing unit 12, and generates buried object information D2 including information such as the type, shape, position and / or size of the identified buried object (S4).

[0065] The buried object identifying unit 13 then outputs the generated buried object information D2 (S5). As a result of the above, the buried object discrimination result is output based on the buried object information D2, and this series of steps in the buried object discrimination method is completed.

[0066] (1-3) Advantages of the Present Embodiment If the relative permittivity used in the synthetic aperture processing is not the correct value, as disclosed in the above-mentioned Patent Document 1, the image that should be converged is over-processed, resulting in distortion of the image. Image distortion results in an apparent expansion of the area where buried objects are likely to exist, which reduces the accuracy of the location of the identified buried object. Furthermore, the buried object identification unit may erroneously detect the excessive distortion as a false image, resulting in overdetection. Therefore, if the buried object identification unit 13 uses underground exploration data D1 that has been subjected to synthetic aperture processing without using the correct relative permittivity to identify buried objects, this will undermine the reliability of the buried object discrimination device.

[0067] In this regard, the buried object discrimination device 2 of this embodiment integrally includes a dielectric constant estimation unit 11 that estimates the dielectric constant of the underground medium based on the underground exploration data D1, and an aperture synthesis processing unit 12 that performs aperture synthesis processing on the same underground exploration data D1 using the estimated dielectric constant E1 estimated by the dielectric constant estimation unit 11, and therefore can ensure the reproducibility of the aperture synthesis processed underground exploration data D1 (aperture synthesis exploration data D3) that is input to the buried object identification unit 13. As a result, the buried object discrimination device 2 can improve the buried object discrimination performance of the buried object identification unit 13, and thus can discriminate buried objects with high reliability.

[0068] 8, in which the same reference numerals are assigned to parts corresponding to those in FIGS. 1A and 1B, shows a buried object management system 20 according to a second embodiment. This buried object management system 20 is configured in the same manner as the buried object management system 1 according to the first embodiment, except that a buried object visualization system 21 is connected to the buried object discrimination device 2 instead of the information display device 3.

[0069] The buried object visualization system 21 comprises a buried object information storage device 22, a map information storage device 23, and an information display device 24. The buried object information storage device 22 is composed of a large-capacity nonvolatile storage device such as an external hard disk drive. The buried object information storage device 22 sequentially stores and accumulates the buried object information D2 output from the buried object discrimination device 2.

[0070] The map information storage device 23 is also configured as a large-capacity nonvolatile storage device such as an external hard disk drive, etc. The map information storage device 23 stores in advance at least map information of the search area.

[0071] The information display device 24 is configured as an information processing device equipped with a liquid crystal display, an organic EL display, etc. Based on information such as the type, shape, location and / or size of buried objects contained in the buried object information group D2A stored in the buried object information accumulation device 22 and the map information DM of the search area stored in the map information storage device 23, the information display device 24 generates and displays an underground search result screen 30 as shown in Fig. 9 which integrates this information.

[0072] The underground exploration result screen 30 comprises a map display area 30A and a buried object information display area 30B. The map display area 30A displays a map 30AA with an image 30AB of the buried object superimposed thereon, and the buried object information display area 30B displays a list 30BA listing the type, material, and / or size of the buried object.

[0073] According to the buried object management system 20 of this embodiment having the above-mentioned configuration, in addition to the effects obtained by the first embodiment, it is possible to obtain the effect that a buried object information group D2A based on underground exploration data D1 acquired at various locations and dates and times can be managed by a single buried object visualization system 21, and various information about buried objects based on the buried object information group D2A can be presented to the user in a visually easy-to-understand manner.

[0074] (3) Third Embodiment Figure 10, in which the same reference numerals are assigned to parts corresponding to those in Figure 1B, shows a buried object management system 40 according to a third embodiment. This buried object management system 40 differs significantly from the buried object management system 1 of the second embodiment in that a buried object discrimination device 41 according to claim 2 shown in Figure 11 is provided with a depth conversion unit 42. The depth conversion unit 42 is a functional unit realized by the aforementioned CPU included in the buried object discrimination device 41 executing a corresponding program stored in the aforementioned memory included in the buried object discrimination device 41.

[0075] The depth conversion unit 42 is notified by the dielectric constant estimation unit 11 of the estimated dielectric constant E1 estimated by the dielectric constant estimation unit 11. The depth conversion unit 42 then refers to the estimated dielectric constant E1 based on information about the position of the buried object on the time axis included in the buried object information D2 provided by the buried object identification unit 13 (information about the time from when the underground radar device emits electromagnetic waves into the ground until the electromagnetic waves are reflected by the buried object and return to the underground radar device), calculates the distance from the ground surface of the buried object (hereinafter referred to as the depth or depth position), and outputs depth-attached buried object information D2' including the calculated depth information. The depth-attached buried object information D2' is output to the information display device 3, and a predetermined exploration result screen showing information such as the type, shape, position, and size of the buried object included in the depth-attached buried object information D2' is displayed on the information display device 3.

[0076] In this case, if there are multiple estimated relative dielectric constants E1, the depth conversion unit 42 uses a portion of the estimated relative dielectric constant E1 estimated at the position closest to the position of the buried object included in the buried object information D2 of the buried object identified by the buried object identification unit 13.

[0077] Furthermore, as shown in Figure 3, when a reaction (clear point) originating from a buried object appears in multiple two-dimensional cross sections within the underground exploration data D1 for one buried object, the dielectric constant estimation unit 43 may estimate different dielectric constants for each cross section of the same buried object (i.e., estimate multiple dielectric constants for the same buried object).

[0078] In such a case, the depth conversion unit 42 may extract a range of the underground exploration data D1 in which it can be determined that the same buried object has been identified based on the buried object information D2 of the buried object identified by the buried object identification unit 13, and may extract multiple relative dielectric constants to be used for depth conversion from the estimated relative dielectric constant E1 within that range, and may adopt, for example, the average value of the extracted relative dielectric constants as the relative dielectric constant to be used for converting the buried object to its depth.

[0079] Furthermore, if the size of the buried object included in the buried object information D2 is equal to or larger than a specified size, the range of the underground exploration data D1 may be specified for each part of the buried object, and a representative value of the estimated relative dielectric constant E1 within the specified range may be calculated based on the relative dielectric constant estimated by the relative dielectric constant estimation unit 43 within the range, and used by the depth conversion unit 42.

[0080] Next, we will explain the method of converting the depth of a buried object in the depth conversion unit 42. As shown in Fig. 2(A), the underground exploration data D1 has at least axes of the exploration progress direction (x direction in Fig. 2(A)) and the time direction (t direction in Fig. 2(A)). The time t here is the elapsed time from when the underground radar device emits electromagnetic waves into the ground to when the underground radar device receives the electromagnetic waves reflected from the ground.

[0081] For this reason, the buried object identification unit 13 normally identifies the position of a buried object on the xt plane. Of these, the position in the time axis direction is sufficient to know the relative positional relationship of the depths of multiple identified buried objects, but it does not represent the absolute position of the depth within the area where the search was performed.

[0082] To determine the absolute depth position of a buried object, it is necessary to determine the speed v at which the electromagnetic waves emitted from the underground radar device travel underground and convert it based on the position t of the buried object on the time axis.

[0083] Here, the conversion from time t to depth z is, for example, This can be done by

[0084] Furthermore, the speed v at which electromagnetic waves travel underground is calculated by multiplying the speed of light c in a vacuum by the relative dielectric constant ε γ Using the following equation It can be expressed as follows.

[0085] Therefore, the depth conversion unit 42 converts the position of the buried object on the time axis included in the buried object information D2 into a depth using equations (1) and (2) and the estimated relative dielectric constant E1 estimated by the relative dielectric constant estimation unit 11, and outputs buried object information with depth D2' including the converted depth information. The information display device 3 then displays the buried object information with depth D2'.

[0086] An example of the depth-added buried object information D2' displayed on the information display device 3 will be described with reference to Figure 7. Figure 7(A) is an example of buried object information for two-dimensional underground exploration data. In the depth-added buried object information D2', the position is represented by the position and depth relative to the exploration direction. A point object may be represented by a single point, and a line object may be represented by a line segment connecting the start point and end point. Figure 7(B) is an example of the depth-added buried object information D2' for three-dimensional underground exploration data. In this case, the depth-added buried object information D2' also has a component perpendicular to the exploration direction.

[0087] According to the buried object management system 40 of this embodiment having the above-mentioned configuration, the absolute depthwise position of the buried object within the search area is displayed on the information display device 3, so in addition to the effect obtained by the first embodiment, the user can also intuitively confirm the depth position of the buried object.

[0088] Furthermore, instead of the information display device 3 connected to the buried object discrimination device 41 shown in Figure 11, the buried object visualization system 21 shown in Figure 8 is connected, and depth-specific buried object information D2' is stored in the buried object information storage device 22. Based on the stored buried object information group D2A and the map information DM of the search area stored in the map information storage device 23, an underground search result screen 44 such as that shown in Figure 12 is generated and displayed.

[0089] The underground exploration result screen 44 comprises a map display area 44A and a buried object information display area 44B. The map display area 44A displays three-dimensionally rendered images 44AA and 44AB of the exploration surface and buried objects, along with numerical values ​​representing the vertical distances of these buried objects from the exploration surface. The buried object information display area 44B displays a list 44BA listing the type, material, and / or size of the buried objects.

[0090] According to the buried object management system having the above configuration, the absolute depthwise position of buried objects within the search area, as well as depth-specific buried object information D2' generated based on multiple underground search data D1, are superimposed on the same map information DM for management, thereby enabling the user to intuitively confirm the buried position between buried objects.

[0091] (4) Fourth Embodiment Figure 15, in which the same reference numerals are assigned to parts corresponding to those in Figure 10, shows a buried object management system 60 according to a fourth embodiment. This buried object management system 60 differs from the buried object management system 40 of the third embodiment in that, in the buried object discrimination device 61 according to claim 3 shown in Figure 14, the buried object identification unit 13 outputs the buried object information D2 to the relative dielectric constant estimation unit 53 in addition to the depth conversion unit 42, the relative dielectric constant estimation unit 53 re-estimates the relative dielectric constant of the underground medium as necessary based on the buried object information D2 and the underground exploration data D1, and the aperture synthesis processing unit 54 can re-execute aperture synthesis processing based on the estimated relative dielectric constant E1 obtained by the re-estimation and an aperture synthesis processing re-execution flag FM input from outside the buried object discrimination device 61.

[0092] 13 shows the flow of the buried object discrimination process performed by the buried object discriminator 61 of this embodiment for discriminating buried objects. This buried object discrimination process starts when the buried object discriminator 61 is provided with the underground exploration data D1.

[0093] The dielectric constant estimating unit 53 estimates the dielectric constant of the underground medium based on the input underground exploration data D1, and outputs the estimated dielectric constant E1 to the aperture synthesis processing unit 54 (S11). The aperture synthesis processing unit 54 also performs aperture synthesis processing on the input underground exploration data D1 using the estimated dielectric constant E1 provided by the dielectric constant estimating unit 53, and outputs the above-mentioned synthetic aperture exploration data D3 thus obtained to the buried object identifying unit 13 (S12).

[0094] The buried object identification unit 13 identifies buried objects based on the input underground exploration data D1 and the synthetic aperture exploration data D3 provided by the synthetic aperture processing unit 54, generates buried object information D2 including information such as the type, shape, position and / or size of the identified buried object, and outputs it to the depth conversion unit 42 (S13).

[0095] The depth conversion unit 42 compares the position where the estimated relative dielectric constant E1 is estimated with the position of the buried object included in the buried object information D2 based on the position of the buried object included in the buried object information D2 provided by the buried object identification unit 52, and calculates the distance between the position of the relative dielectric constant E1 estimated by the relative dielectric constant estimation unit 53 and the position of the buried object included in the buried object information D2 (S14). Then, it is determined whether this distance is within an allowable range (S15).

[0096] Obtaining a positive result in this determination means that there is a high possibility that the estimated relative dielectric constant E1 estimated by the relative dielectric constant estimating unit 53 is accurate. Thus, at this time, the buried object identifying unit 52 does not output the buried object information D2 generated at that time to the relative dielectric constant estimating unit 53.

[0097] As a result, the depth conversion unit 42 converts the position of the buried object on the time axis into a depth based on the buried object information D2 from the buried object identification unit 52 and the estimated relative dielectric constant E1 provided by the relative dielectric constant estimation unit 53 (S18).

[0098] The depth conversion unit 42 also generates depth-added buried object information D2' including the converted depth position and outputs the generated buried object information D2' (S19). With this, the buried object discrimination process is completed.

[0099] In contrast, a negative result in the determination at step S15 means that there is a possibility that the estimated relative dielectric constant E1 estimated by the relative dielectric constant estimation unit 53 is not accurate. Thus, at this time, the buried object identification unit 52 outputs the buried object information D2 generated at that time to the relative dielectric constant estimation unit 53.

[0100] The relative dielectric constant estimating unit 53 re-estimates the relative dielectric constant of the underground medium based on the buried object information D2 provided by the buried object identifying unit and the underground exploration data D1 (S16).

[0101] Specifically, the dielectric constant estimation unit 53 re-estimates the dielectric constant by performing a fitting process in which the parameters are changed to fit the logical arrival time to the underground exploration data D1, using the position of the buried object identified by the buried object identification unit 52 as a substitute for the position of the above-mentioned clear point.

[0102] The aperture synthesis processing unit 54 then determines whether the aperture synthesis processing re-execution flag FM indicates 1, which prompts re-execution (S17). A negative result in the determination of step S17 means that the user has determined that re-execution of aperture synthesis processing is unnecessary, so steps S18 and S19 are executed, and the processing ends. A positive result in the determination of step S17 means that re-execution of aperture synthesis processing is necessary. In this case, the aperture synthesis processing re-execution flag FM is changed to 0 (S20), and then feedback is given to step S12 to re-execute aperture synthesis processing. The subsequent processing is as described above.

[0103] As a result, the depth conversion unit 42 converts the position of the buried object on the time axis into a depth based on the buried object information D2 from the buried object identification unit 52 and the re-estimated value of the dielectric constant given by the dielectric constant estimation unit 53 (S18).

[0104] The depth conversion unit 42 also generates buried object information D2' including the converted depth and outputs the generated buried object information D2' (S19). This completes the buried object discrimination process.

[0105] Even if the estimated relative permittivity E1 corresponding to the buried object information has already been estimated, the relative permittivity estimation unit 53 may be configured to output a negative result in step S15 so that it re-estimates the estimated relative permittivity E1 by using the buried object information D2 in addition to the underground exploration data D1. In this case, if the buried object information D2 of the buried object identified by the buried object identification unit 13 includes the depth position, horizontal position, type, shape, and size of the buried object, in step S16, the relative permittivity estimation unit 53 estimates the relative permittivity using a logical arrival time that is assumed based on parameters such as the depth position, horizontal position, type, shape, and size of the buried object and that may be more valid than the logical arrival time assumed in the first estimation.

[0106] Here, in the above-mentioned step S11, if the relative permittivity estimation unit 53 performs a fitting method in which the clear points described in Non-Patent Document 2 are assumed to be reflection points of electromagnetic waves in buried objects, there is a risk that unclear reflection images actually measured in non-metallic buried objects, etc. will be overlooked.

[0107] In this regard, in the buried object discrimination device 61 of the present embodiment, when the position of the estimated relative dielectric constant E1 estimated by the relative dielectric constant estimating unit 53 and the position of the buried object identified by the buried object identifying unit 13 are far apart, the relative dielectric constant estimating unit 53 re-estimates the estimated relative dielectric constant E1 using the position information of the buried object identified by the buried object identifying unit 13, and the position of the buried object on the time axis is converted to depth based on the estimated relative dielectric constant E1 obtained by this re-estimation.Therefore, even for reflection images that are missed when it is assumed that clear points are reflection points of buried objects, depth conversion can be performed using the re-estimated estimated relative dielectric constant E1, and the accuracy of the converted depth of the buried object can be improved.

[0108] Furthermore, if the relative dielectric constant used in the aperture synthesis process is not the correct value, the image that should be aggregated by the aperture synthesis process will be over-processed, resulting in distortion of the image. Therefore, if the underground exploration data D1 (synthetic aperture exploration data D3) that has been subjected to the aperture synthesis process using the incorrect relative dielectric constant is used to perform identification in the buried object identification unit 13, this can lead to overdetection and other problems.

[0109] In this regard, in the buried object discrimination device 61 of this embodiment, aperture synthesis processing is performed again using the dielectric constant re-estimated by the dielectric constant estimation unit 53, so that aperture synthesis processing using an appropriate dielectric constant can be performed on the underground exploration data D1.

[0110] Therefore, according to this embodiment, the accuracy of identifying buried objects in the buried object identifying unit 13 can be further improved, and thus it is possible to obtain the effect of being able to distinguish buried objects with even higher reliability.

[0111] (5) Fifth Embodiment Figure 17, in which the same reference numerals are assigned to parts corresponding to those in Figure 15, shows the configuration of a buried object management system 70 according to a fifth embodiment. The buried object management system 70 in Figure 17 differs greatly from the buried object management system 61 according to the fourth embodiment in that a selection unit 75 is provided in comparison with the buried object discrimination device 61, like the buried object discrimination device 71 according to claim 4 shown in Figure 16(A).

[0112] 17 is a functional unit realized by the CPU of the buried object discrimination device 71 executing a corresponding program stored in the memory of the buried object discrimination device 71. The selection unit 75 has a function of outputting a depth conversion re-execution flag F3, an aperture synthesis processing re-execution flag FM, and a buried object identification re-execution flag F2 which control whether or not to re-execute aperture synthesis processing in the aperture synthesis processing unit 54, whether or not to re-execute buried object identification in the buried object identification unit 64, and whether or not to re-execute depth conversion in the depth conversion unit 62, based on the estimated relative dielectric constant E1 estimated in the relative dielectric constant estimation unit 53, the underground exploration data D1, and the synthetic aperture exploration data D3.

[0113] Specifically, the selection unit 75 will be described with reference to the block diagram shown in Figure 16(B). The estimated relative dielectric constant E1 input to the selection unit 75 is input to a theoretical underground exploration data generation unit 76 and a depth conversion re-execution flag generation unit 79. The theoretical underground exploration data generation unit 76 assumes that a buried object exists at the position where the estimated relative dielectric constant E1 is estimated, generates a theoretical reflection image based on the estimated relative dielectric constant E1, and outputs it as theoretical underground exploration data D1T to an exploration data agreement calculation unit 77. The theoretical underground exploration data D1T may be generated by performing a numerical calculation simulation using the finite element method, the finite difference time domain method, or the like.

[0114] The exploration data match rate calculation unit 77 calculates the exploration data match rate V1, which represents the match rate of the exploration data, based on the underground exploration data D1 and the theoretical underground exploration data D1T. If the underground exploration data D1 and the theoretical underground exploration data D1T are configured with arrays of the same size, the match rate can be calculated, for example, as the average of the absolute values ​​of their differences. A high exploration data match rate indicates that the estimated relative dielectric constant E1 estimated by the relative dielectric constant estimation unit 53 can well explain the underground exploration data D1, and can therefore be used to evaluate the validity of the estimated relative dielectric constant E1. The calculated exploration data match rate V1 is output to the depth conversion re-execution flag generation unit 79 and the aperture synthesis processing re-execution flag generation unit 710.

[0115] The exploration data convergence rate calculation unit 78 calculates the convergence rate as the exploration data convergence rate V2 based on the synthetic aperture exploration data D3. The method of calculating the convergence rate can be, for example, the method disclosed in Non-Patent Document 4. Since the synthetic aperture processing is a process for correcting distortion of the reflected image, the exploration data convergence rate V2, which indicates the convergence rate, can be used to evaluate the validity of the synthetic aperture processing performed by the synthetic aperture processing unit 54. The exploration data convergence rate V2 is output to the buried object identification re-execution flag generation unit 711.

[0116] The depth conversion retry flag generation unit 79 holds the estimated relative dielectric constant E1, the exploration data agreement rate V1, or both, and generates a depth conversion retry flag F3 based on changes in the held estimated relative dielectric constant E1, the exploration data agreement rate V1, or both.

[0117] The synthetic aperture processing re-execution flag generation unit 710 holds the search data match rate V1 and generates a synthetic aperture processing re-execution flag FM based on changes in the held search data match rate V1.

[0118] The buried object identification retry flag generating unit 711 holds the search data convergence rate V2 and generates a buried object identification retry flag F2 based on changes in the held search data convergence rate V2.

[0119] Some of the patterns for generating the depth conversion re-execution flag F3, the aperture synthesis processing re-execution flag FM, and the buried object identification re-execution flag F2 will be described based on the specific example shown in FIG.

[0120] As shown in FIG. 18 , if the difference between the first and second estimated relative dielectric constants E1 is equal to or greater than a predetermined threshold, the depth conversion process should be retried because a significant difference in the conversion results from the depth conversion unit 62 is likely to occur. Therefore, the flag state is set to “1” (No. 1). Conversely, if the difference is equal to or less than the predetermined threshold, a retried depth conversion process is likely to result in no significant difference in depth. Therefore, the flag state is set to “0” (No. 2). Furthermore, if the difference between the first and second search data match rates V1 is positive, the depth conversion process should be retried because a more reasonable estimated relative dielectric constant E1 is likely to be obtained. Therefore, the flag state is set to “1” (No. 3). Conversely, if the difference between the first and second search data match rates V1 is negative, the depth conversion process should be retried because a less reasonable estimated relative dielectric constant E1 is likely to be obtained. Therefore, the flag state is set to “0” (No. 4).

[0121] The aperture synthesis processing re-execution flag FM is also set to "1" (No. 5), indicating that aperture synthesis processing should be re-executed, if the difference between the first and second numerical values ​​of the search data match rate V1 is positive, since it is considered that a more appropriate estimated permittivity E1 has been obtained. Conversely, if the difference between the first and second numerical values ​​of the search data match rate V1 is negative, it is considered that a less appropriate estimated permittivity E1 has been obtained, so it is set to "0" (No. 6), indicating that aperture synthesis processing should not be re-executed.

[0122] The buried object identification retry flag F2 is also set to "1" (No. 7), indicating that buried object identification processing should be retried, if the difference between the first and second values ​​of the exploration data convergence rate V2 is positive, since it is considered that more appropriate aperture synthetic exploration data D3 has been obtained. Conversely, if the difference between the first and second values ​​of the exploration data convergence rate V2 is negative, it is considered that less appropriate aperture synthetic exploration data D3 has been obtained, and it is set to "0" (No. 8), indicating that buried object identification processing should not be retried.

[0123] The selection unit 75 may determine the validity of the estimated relative dielectric constant E1, the exploration data match rate V1, and the exploration data convergence rate V2 before and after the re-execution by comparing the magnitudes before and after the re-execution as described above, or may determine the validity based on whether a predetermined threshold value has been reached or whether the rate of change has exceeded a predetermined threshold value.

[0124] The buried object management system 70 of this embodiment having the above configuration can regenerate the synthetic aperture exploration data D3 to be input to the buried object identification unit 64 based on the high-level buried object identification results in the buried object identification unit 64. This improves the quality of the synthetic aperture exploration data D3 to be input to the buried object identification unit 64, and can improve the buried object identification accuracy in the buried object identification unit 64.

[0125] Note that the buried object identification result (buried object information D2) that uses updated synthetic aperture exploration data D3 as input may be updated, and therefore this identification result may be returned to the dielectric constant estimation unit 53, and the dielectric constant re-estimation in the dielectric constant estimation unit 53, the aperture synthesis processing in the synthetic aperture processing unit 54, and the above-mentioned selection action in the selection unit 75 may be performed multiple times. By doing so, the quality of the synthetic aperture exploration data D3 input to the buried object identification unit 64 is improved, and the identification accuracy of the buried object identification unit 64 can be improved.

[0126] In this case, for example, among the data shown in FIG. 18 , each time the dielectric constant is re-estimated by the dielectric constant estimating unit 53, the new dielectric constant may be overwritten as the initial estimated value of the dielectric constant and held, or data for a plurality of times may be held.

[0127] By doing so, each time the dielectric constant is re-estimated by the dielectric constant estimating unit 53, the quality of the re-estimated dielectric constant and the synthetic aperture exploration data D3 generated by the synthetic aperture processing unit 54 based on the re-estimated dielectric constant improves, thereby further improving the accuracy of identifying buried objects in the buried object identifying unit 64. Therefore, the buried object management system 70 of this embodiment can more reliably identify buried objects.

[0128] (6) Sixth Embodiment Figure 19, in which the same reference numerals are assigned to parts corresponding to those in Figure 16(B), shows the configuration of a selection unit 75' of a buried object management system according to a sixth embodiment. The selection unit 75' of Figure 19 differs significantly from the selection unit 75 according to the fifth embodiment in that it can output a search data match rate V1 and a search data convergence rate V2.

[0129] The exploration data match rate V1 and exploration data convergence rate V2 output from the selection unit 75' are values ​​that evaluate the validity of the relative dielectric constant estimation process by the relative dielectric constant estimation unit 53 and the aperture synthesis process by the aperture synthesis processing unit 54, respectively, and in the sixth embodiment, these exploration data match rate V1 and exploration data convergence rate V2 are output outside the buried object discrimination device.

[0130] By doing so, the information display device 3 connected to the buried object discrimination device can display the search data match rate V1 and the search data convergence rate V2, allowing the user to confirm the validity of the buried object identification results obtained by the buried object discrimination device. Therefore, the buried object management system of this embodiment can more reliably identify buried objects.

[0131] (7) Seventh Embodiment Figure 21, in which parts corresponding to those in Figure 1B are assigned the same reference numerals, shows a buried object management system 90 according to a seventh embodiment. This buried object management system 90 differs from the first embodiment in that a buried object discrimination device 91 in Figure 20(A) has a buried object learning model generation unit 97 shown in Figure 20(B), the buried object learning model generation unit 97 generates a buried object learning model LM using acquired underground exploration data D1, aperture synthetic exploration data D3 generated within the device itself, and buried object presence information A1 including information indicating at least the presence or absence of a buried object in the exploration area in which the underground exploration data D1 was acquired, and the buried object management system 90 identifies buried objects using the generated buried object learning model LM.

[0132] In practice, in this embodiment, the learning data generation unit 95 generates learning data DL based on the underground exploration data D1, the synthetic aperture exploration data D3 generated based on the underground exploration data D1, and the buried object presence information A1, and stores the learning data DL in the learning data storage unit 94. Here, in Fig. 20(B) the learning data storage unit 94 is shown as being included in the buried object learning model generation unit 97, but it may be located inside or outside the buried object discrimination device 91 of the buried object learning model generation unit 97.

[0133] The buried object learning model generation unit 97 then uses the learning data group DLA stored in memory as described above to learn the type, shape, position, size, etc. of buried objects in the underground exploration data D1 and / or the synthetic aperture exploration data D3 using the learner 96, thereby generating a buried object learning model LM. This buried object learning model LM is an AI (Artificial Intelligence) model configured, for example, by a neural network.

[0134] Then, when the buried object identification unit 93 is given underground exploration data D1 acquired through exploration work in a new exploration area, etc., and / or aperture synthesis exploration data D3 from the aperture synthesis processing unit 12, it identifies the type, shape, position, size, etc. of the buried object based on the buried object learning model LM, and outputs buried object information D2 based on the identification results to the information display device 3.

[0135] The buried object learning model generation unit 97 then uses the continuously accumulated learning data group DLA to re-learn the type, shape, position, size, etc. of buried objects in the underground exploration data D1 and / or the synthetic aperture exploration data D3 in the learning device 96, and updates the buried object learning model LM as necessary.

[0136] According to the buried object management system 90 of this embodiment having the above-mentioned configuration, the synthetic aperture exploration data D3 corresponding to the underground exploration data D1 is generated within the buried object discrimination device 91, so that the underground exploration data D1 and / or the synthetic aperture exploration data D3 can be automatically associated with buried object information A1 such as the type, shape, position, and size of the buried object within the buried object discrimination device 91.

[0137] Therefore, compared to the conventional case where the synthetic aperture exploration data D3 is generated outside the buried object discrimination device 91, it is possible to reduce the cost and effort required to match the underground exploration data D1 and / or synthetic aperture exploration data D3 with buried object information A1 such as the type, shape, position and size of the buried object.

[0138] Furthermore, in the buried object management system 90, the buried object learning model LM is updated as needed as described above, so the buried object identification performance of the buried object identification unit 93 can be continuously improved.

[0139] As a result, according to the buried object management system 90 of this embodiment, in addition to the effect obtained by the first embodiment, the effect of being able to construct a buried object discrimination device with high buried object discrimination performance at low cost can be obtained.

[0140] Furthermore, in this embodiment, a buried object learning model generation unit 97' according to claim 7 shown in Figure 22 is provided instead of the buried object learning model generation unit 97, so that the underground exploration data D1 and the aperture synthesis data D3 are constantly stored in the exploration data storage unit 166, and by inputting buried object presence information A1 within the exploration area, the learning data generation unit 95 generates learning data DL using the accumulated exploration data group DPL.

[0141] As a result, according to the buried object management system 90 of this embodiment, even if buried object presence information for the exploration area is acquired at a time different from the time when underground exploration data is acquired, the learning data DL can be automatically associated and appropriate learning data DL can be accumulated within the buried object management system 90. Furthermore, by appropriately updating the buried object learning model LM, it is possible to obtain the effect that the buried object management system 90 can be operated while continuously improving the buried object identification performance of the buried object identification unit 93.

[0142] (8) Eighth Embodiment Figure 24, in which parts corresponding to those in Figures 1A and 1B are assigned the same reference numerals, shows a buried object management system 100 according to an eighth embodiment. As shown in Figure 23, which shows a buried object discrimination device 101 according to claim 8, this buried object management system 100 differs significantly from the buried object management system 1 of the first embodiment in that a data compression unit 102 that compresses input underground exploration data D1 is additionally provided in the buried object discrimination device 2 according to claim 1 shown in Figure 1A.

[0143] In practice, the data compression unit 102 compresses the input underground exploration data D1 by extracting data of sparse points, for example, on the x-axis or t-axis, or on both the x-axis and t-axis, as shown in Fig. 2A. However, various other compression methods, such as reducing the number of data bits at each point, can also be used to compress the underground exploration data D1 by the data compression unit 102.

[0144] The data compression unit 102 then outputs the compressed underground exploration data D1 as compressed underground exploration data D10 to the relative dielectric constant estimation unit 11, the aperture synthesis processing unit 12, and the buried object identification unit 13. As a result, the subsequent processes in the relative dielectric constant estimation unit 11, the aperture synthesis processing unit 12, and the buried object identification unit 13 are performed based on the compressed underground exploration data D10.

[0145] According to the buried object management system 100 of this embodiment having the above-mentioned configuration, in addition to the effects obtained by the first embodiment, it is possible to obtain the effect of reducing the calculation costs of each process in the relative dielectric constant estimation unit 11, the aperture synthesis processing unit 12, and the buried object identification unit 13 of the buried object discrimination device 101.

[0146] Furthermore, according to this buried object management system 100, it is possible to reduce the memory capacity required for various calculation processes in the buried object discrimination device 101, which has the effect of enabling buried objects to be discriminated at lower cost and in a shorter time.

[0147] 1A and 1B, the same reference numerals are assigned to parts corresponding to those in the ninth embodiment, and Fig. 26 shows a buried object management system 110 according to the ninth embodiment. As shown in Fig. 25, which shows a buried object discrimination device 111 of claim 9, this buried object management system 110 differs significantly from the buried object management system 1 of the first embodiment in that a resizing processing unit 112 that resizes the underground exploration data D1 and the synthetic aperture exploration data D3 is additionally provided in the buried object discrimination device 111.

[0148] In practice, the resizing processing unit 112 is provided with the input underground exploration data D1 and the synthetic aperture exploration data D3 output from the aperture synthesis processing unit 12. The resizing processing unit 112 then performs resizing processing on the underground exploration data D1 and the synthetic aperture exploration data D3, and outputs the resized underground exploration data D1 and resized synthetic aperture exploration data D3' to the buried object identification unit 13.

[0149] The term "resizing" used here refers to increasing or decreasing the number of data points arranged along the x-axis or t-axis directions in Fig. 2A. The resizing unit presets the data size after resizing, such as 500 x 1000 for two-dimensional data such as that shown in Fig. 2A.

[0150] The resizing processing unit 112 then resizes the input underground exploration data D1 and synthetic aperture exploration data D3 to the specified data size. Note that the resized data sizes of the underground exploration data D1 and synthetic aperture exploration data D3 may be the same or different.

[0151] As a method for changing the number of data points in such resizing processing, interpolation methods such as nearest neighbor interpolation, linear interpolation, Lagrange interpolation, centroid interpolation, and quadratic spline interpolation can be used.

[0152] According to the buried object management system 110 of this embodiment having the above-mentioned configuration, the size of the data that flows to subsequent processing can be kept constant within the buried object discrimination device 111 by such resizing processing, so that even if the size of the underground exploration data D1 acquired and generated by underground radar devices of different manufacturers or different models changes, for example, it is possible to reuse some or all of the data.

[0153] Therefore, according to this buried object management system 110, even if the buried object identification unit 13 has a configuration that identifies buried objects using the buried object learning model LM as in the seventh embodiment, there is no need to modify the network size of the buried object learning model LM (Figure 19), and thus it is possible to achieve the effect of constructing a more versatile system.

[0154] (10) Tenth Embodiment Figure 28, in which parts corresponding to those in Figure 21 are assigned the same reference numerals, shows a buried object management system 120 according to a tenth embodiment. This buried object management system 120 is configured in the same manner as the buried object management system 110 of the ninth embodiment, except that a data dividing unit 122 for dividing the underground exploration data D1 is additionally provided in a buried object discrimination device 121 according to claim 10 shown in Figure 27.

[0155] In practice, the data dividing unit 122 is provided with the underground exploration data D1 input to the buried object discrimination device 121. The data dividing unit 122 then performs a dividing process to divide this underground exploration data D1 into a plurality of pieces, and outputs the divided underground exploration data (hereinafter referred to as divided underground exploration data) D11 to the relative dielectric constant estimation unit 11, the aperture synthesis processing unit 12, and the resizing processing unit 112 in sequence.

[0156] In this case, the underground exploration data D1 is divided, for example, in the exploration direction (x-axis direction in FIG. 2A), into a plurality of divided underground exploration data D11 corresponding to a predetermined distance in real space. For example, if the underground radar device travels 100 m in the exploration direction, the data dividing unit 122 divides the underground exploration data D1 into 10 divided underground exploration data D11, each with a travel distance of 10 m in the exploration direction.

[0157] According to the buried object management system 120 of this embodiment having the above-mentioned configuration, it is possible to reduce the calculation costs of the relative dielectric constant estimation unit 11, the aperture synthesis processing unit 12, and the buried object identification unit 13 in the buried object discrimination device 121 without incurring a reduction in the amount of information due to data compression as in the ninth embodiment, for example.

[0158] Furthermore, according to the buried object management system 120, it is possible to reduce the memory capacity of the buried object discrimination device 121, and it is possible to obtain the effect that the buried object discrimination process can be performed at lower cost and in a shorter time.

[0159] (11) Eleventh Embodiment There are several types of underground radar devices with different transmitting and receiving antenna structures. Figures 29(A-1) and (B-1) show typical structural examples of such transmitting and receiving antennas. Figure 29(A-1) shows an antenna structure configured as a transmitting and receiving antenna 130 in which the transmitting antenna Tx and the receiving antenna Rx are integrated, while Figure 29(B-1) shows an antenna structure of a transmitting and receiving antenna 131 in which the transmitting antenna Tx and the receiving antenna Rx are provided separately.

[0160] In the case of the antenna structure of Fig. 29(A-1), as shown in Fig. 29(A-2), most of the electromagnetic waves emitted from the transmitting / receiving antenna 130 toward the ground are reflected by a buried object 132 buried directly below, and the reflected waves are received by the transmitting / receiving antenna 130. In the case of the antenna structure of Fig. 29(B-1), as shown in Fig. 29(B-2), of the waves emitted from the transmitting antenna Tx of the transmitting / receiving antenna 131 toward the ground and reflected by the buried object 132 directly below, only the reflected waves of a portion of the electromagnetic waves that are incident on the buried object 132 at a certain angle of incidence are received by the receiving antenna Rx of the transmitting / receiving antenna 131.

[0161] As is clear from a comparison of Figures 29(A-2) and 29(B-2), the path taken by the electromagnetic waves emitted from the transmitting and receiving antennas 130, 131 and reflected by the buried object 132 before being received by the transmitting and receiving antennas 130, 131 varies depending on the antenna structure of the underground radar device, and accordingly the time taken from when the transmitting and receiving antennas 130, 131 emit the electromagnetic waves to when the reflected waves from the buried object 132 are received by the transmitting and receiving antennas 130, 131 also varies. With regard to this reception time, if the horizontal axis represents the position of the underground radar device and the vertical axis represents the time required for reception, it can be expressed as the entire edges 133 and 134 in Figures 29(A-4) and (B-4).

[0162] For example, when the antenna structure of the underground radar device is as shown in FIG. 29(A-1), the x coordinate of the position directly above the buried object 132 in the transmitting / receiving antenna 130 of the underground radar device is expressed as x ０ The time it takes for the electromagnetic wave emitted from that position to be reflected by the buried object 132 and return to the transmitting / receiving antenna 130 is t ０ , the x-coordinate of the transmitting / receiving antenna 130 is x ｉ The time it takes for the electromagnetic wave emitted from the position to be reflected by the buried object 132 and return to the transmitting / receiving antenna 130 is t ｉ , the speed at which the electromagnetic wave travels underground is v, and time t ｉ The radar reflection formula is as follows: It is defined as follows:

[0163] In addition, when the antenna structure of the underground radar device is as shown in FIG. 29(B-1), as shown in FIG. 29(B-3), the x coordinate of the position directly above the buried object 132 in the transmitting antenna Tx of the transmitting / receiving antenna 131 is expressed as x ０ The time it takes for the electromagnetic wave emitted from that position to be reflected by the buried object 132 and reach the receiving antenna Rx is t ０ , the x coordinate of the transmitting antenna Tx is x ｉ The time it takes for the electromagnetic wave emitted from the position to be reflected by the buried object 132 and reach the receiving antenna Rx is t ｉ , the speed at which the electromagnetic wave travels underground is v, and the distance between the transmitting antenna Tx and the receiving antenna Rx in the transmitting / receiving antenna 131 is TR offset (See FIG. 29(B-1)), then time tｉ The radar reflection formula is as follows: It is defined as follows:

[0164] In this case, for example, in the buried object discrimination device 2 of the first embodiment, the relative dielectric constant estimation unit 11 and the aperture synthesis processing unit 12 perform calculations to estimate the relative dielectric constant and correct distortion of the reflected image by modeling the time it takes for the electromagnetic waves emitted from the underground radar device to be reflected by a buried object and received by the receiving antenna as a radar reflection equation.

[0165] Therefore, if the same radar reflection formula is used to estimate the dielectric constant or perform aperture synthesis processing without distinguishing between the underground exploration data D1 from the underground radar device with the antenna structure of Figure 29 (A-1) and the underground exploration data D1 from the underground radar device with the antenna structure of Figure 29 (B-1), there is a risk of a large error occurring between the estimated dielectric constant and the actual dielectric constant, or distortion occurring in the aperture synthesis image.

[0166] Therefore, in this embodiment, the radar reflection type is switched depending on the antenna structure of the underground radar device, thereby preventing a large error from occurring between the estimated relative dielectric constant and the actual relative dielectric constant and preventing distortion from occurring in the aperture synthesis image.

[0167] 31, in which the same reference numerals are assigned to parts corresponding to those in FIGS. 1A and 1B in relation to the operation of the buried object discrimination device of this embodiment, shows the configuration of a buried object management system 140 according to this embodiment equipped with such functions. This buried object management system 140 differs significantly from the buried object management system 1 of the first embodiment in that the functions of the relative dielectric constant estimation section 142 and the aperture synthesis processing section 143 of the buried object discrimination device 141 are different.

[0168] 32A and 32B show the flow of the buried object discrimination process executed for discriminating a buried object in the buried object discriminating device of this embodiment.

[0169] 32A and 32B, in the buried object discrimination device 141, first, the dielectric constant estimation unit 142 reads the underground exploration data D1 (S501). Then, it is determined whether the underground exploration data D1 includes antenna structure information IA (S502). If the antenna structure information IA is not present, the buried object discrimination process ends after going through the same processes as the buried object management system 1 of the first embodiment (S504, S505, S508 to S510 in FIG. 32B).

[0170] If the underground exploration data D1 includes antenna structure information IA, i.e., if a positive result is obtained in step S502, it is determined whether to change the radar reflection formula R1 of the relative permittivity estimation unit 142 (S511). If a positive result is obtained in step S511, the relative permittivity estimation unit 142 selects the radar reflection formula R1 based on the antenna structure information IA (S503). If a negative result is obtained in step S511, step S503 is skipped. Next, the relative permittivity estimation unit 142 estimates the relative permittivity (S504). The estimated relative permittivity E1 obtained by the relative permittivity estimation unit 142 is input to the aperture synthesis processing unit 143 and read together with the underground exploration data D1 (S505). Next, it is determined whether to change the radar reflection formula R2 of the aperture synthesis processing unit 143 (S506). If a positive result is obtained in step S506, the aperture synthesis processing unit 143 selects the radar reflection formula R2 based on the antenna structure information IA (S507). If a negative result is obtained in step S506, step S507 is skipped. The subsequent processing is the same as that of the buried object management system 1 of the first embodiment (S508 to S510), and then this buried object discrimination processing is completed.

[0171] In this buried object discrimination device 141, a list of antenna structures for each manufacturer and model of underground radar device (hereinafter referred to as the antenna structure list) and radar reflection formulas such as formulas (3) and (4) for each antenna structure are stored and managed in, for example, the above-mentioned memory provided in the buried object discrimination device 141 so that the relative dielectric constant estimation unit 142 and the aperture synthesis processing unit 143 can change the radar reflection formulas R1 and R2.

[0172] According to the buried object management system 140 of this embodiment having the above-described configuration, the accuracy of the dielectric constant estimation by the dielectric constant estimation unit 142 and the aperture synthesis processing by the aperture synthesis processing unit 143 can be kept constant even for underground radar devices with different antenna structures, and the estimated or generated dielectric constant and synthetic aperture exploration data D3 can be kept as homogeneous as possible as a buried object discrimination device 141.

[0173] Therefore, in addition to the effect obtained by the first embodiment, the buried object management system 140 can also provide the effect of ensuring that the output of the buried object identifying unit 13 can be used for general purposes.

[0174] (12) Twelfth embodiment As illustrated in Figures 29(A-3) and 29(B-3), the reflection paths of electromagnetic waves are different for underground radar devices with different antenna structures, and therefore the reflection images from buried objects that appear in the underground exploration data D1 shown in Figure 1B may have unique shapes.

[0175] Therefore, it is believed that the accuracy of identifying buried objects can be improved by storing the given underground exploration data D1 separately for each antenna structure of the underground radar device used to acquire the underground exploration data D1, generating a buried object learning model that identifies buried objects based on the stored underground exploration data D1 for each antenna structure, and identifying buried objects using the generated buried object learning model.

[0176] Figure 34, in which the same reference numerals are assigned to parts corresponding to those in Figure 31, shows a buried object management system 170 of this embodiment that is configured based on this idea. One of the features of this buried object management system 170 is that it has a buried object learning model generation unit 172 shown in Figure 33, generates a buried object learning model LM (LM1, LM2, ...) for each antenna structure of the underground radar device, and identifies buried objects using a corresponding buried object learning model LM from the generated buried object learning models LM.

[0177] In practice, in this embodiment, the antenna structure information IA included in the underground exploration data D1 is held in the learning data DL generated by the learning data generation unit 95 of the buried object learning model generation unit 172. The learner 173 extracts learning data groups DLA having the same type of antenna structure information IA from the accumulated learning data groups DLA, and trains the buried object learning model LM based on these. The buried object learning model generation unit holds a buried object learning model LM (LM1, LM2, ...) for each piece of antenna structure information IA.

[0178] Meanwhile, the buried object identification unit 174 inputs the underground exploration data D1, reads the antenna structure information IA, acquires a suitable buried object learning model LM from the buried object learning model generation unit 172, and identifies buried objects using the acquired buried object learning model LM. The correspondence between the antenna structure information IA and the buried object learning model LM is determined, for example, by referring to a correspondence table stored in a memory provided in this buried object discrimination device.

[0179] The buried object identifying unit 174 then outputs buried object information D2 of the buried object obtained by this identification process from the buried object determining device 171 to the information display device 3. As a result, the buried object identification result is displayed on the information display device 3 based on this buried object information D2.

[0180] According to the buried object management system 170 of this embodiment having the above-described configuration, a buried object learning model LM (LM1, LM2) suitable for each antenna structure is generated within the system for input of underground exploration data D1 acquired by underground radar devices having different structures, and the identification accuracy of the buried object identification unit 174 can be improved while ensuring that the output of the buried object identification unit 174 is of uniform quality.

[0181] (13) Other Embodiments In the first to twelfth embodiments described above, the buried object discrimination device is configured using a single computer device, but the present invention is not limited to this. The buried object discrimination device may also be constructed using multiple computers that make up a distributed computing system.

[0182] Furthermore, in the above-described first to twelfth embodiments, the case has been described in which the relative dielectric constant estimation units 11, 53, 142, the aperture synthesis processing units 12, 54, 63, 73, 143, the buried object identification units 13, 64, 174, the depth conversion units 42, 62, the selection units 75, 75′, the buried object learning model generation units 97, 97′, 172, the data compression unit 102, the data division unit 122, and the resizing processing unit 112 are configured as software, but the present invention is not limited to this, and some or all of these may be configured as hardware.

[0183] Furthermore, in the above-mentioned first to twelfth embodiments, the buried object management systems 1, 20, 40, 60, 70, 90, 110, 120, 140, and 170 have been described as having only the characteristic configurations unique to each of the first to twelfth embodiments, respectively. However, the present invention is not limited to this, and a buried object management system may be constructed by combining some or all of the characteristic configurations unique to each of the first to twelfth embodiments.

[0184] The present invention can be widely applied to buried object discrimination devices of various configurations that discriminate buried objects underground based on underground exploration data obtained by transmitting and receiving electromagnetic waves underground.

[0185] 1, 20, 40, 60, 70, 90, 110, 120, 140, 170... buried object management system, 2, 41, 61, 71, 81, 91, 161, 101, 111, 121, 141, 171... buried object discrimination device, 11, 53, 142... relative dielectric constant estimation unit, 12, 54, 63, 73, 143... aperture synthesis processing unit, 13, 64, 174... buried object identification unit, 30, 44... underground exploration result screen, 42, 62... depth conversion unit, 56, 153... memory, 102... data compression unit, 112... resizing processing unit, 1 22...data division unit, 130...transmitting and receiving antenna, 132...buried object, D1...underground exploration data, D1'...resized underground exploration data, D3'...resized synthetic aperture exploration data, D2, D21, D22...buried object information, D2'...buried object information with depth, D3...synthetic aperture exploration data, D10...compressed underground exploration data, D11...divided underground exploration data, Rx...receiving antenna, Tx...transmitting antenna, D4...synthetic data, E1...estimated relative permittivity, FM...synthetic aperture processing re-execution flag, F2...buried object Re-identification flag, F3...depth conversion re-execution flag, V2...search data convergence rate, V1...search data match rate, D2A...buried object information group, DM...map information, 22...buried object information storage device, 23...map information storage device, 24...information display device, 21...buried object visualization system, 75, 75'...selection unit, A1...buried object presence information, 94...learning data storage unit, 95...learning data generation unit, 166...search data storage unit, DPL...search data group, DLA...learning data group, DL...learning data, I A...antenna structure information, LM, LM1, LM2...buried object learning model, 76...theoretical underground exploration data generation unit, 77...exploration data match rate calculation unit, 78...exploration data convergence rate calculation unit, 79...depth conversion re-execution flag generation unit, 710...aperture synthesis processing re-execution flag generation unit, 711...buried object identification re-execution flag, D1T...theoretical underground exploration data, 97, 97', 172...buried object learning model generation unit, 96, 173...learner, 133, 134...entire edge of buried object reflection signal, R1, R2...radar reflection type.

Claims

1. A buried object discrimination device which discriminates buried objects underground based on underground exploration data obtained by transmitting and receiving electromagnetic waves underground, comprising: a dielectric constant estimation unit which estimates the dielectric constant of an underground medium based on the underground exploration data provided from an external source and outputs an estimated dielectric constant; an aperture synthesis processing unit which uses the estimated dielectric constant to perform aperture synthesis processing on the underground exploration data and outputs synthetic aperture exploration data; and a buried object discrimination unit which performs an identification process to identify the buried object based on the aperture synthesis exploration data and the underground exploration data, and outputs information regarding the buried object as buried object information.

2. The buried object discrimination device according to claim 1, further comprising a depth conversion section which converts the position of the buried object information on the time axis into a depth representing the distance of the buried object from the ground surface based on the buried object information output from the buried object identification section and the estimated relative dielectric constant estimated by the relative dielectric constant estimation section, and outputs buried object information with depth which also includes the converted depth.

3. A buried object discrimination device as defined in claim 2, wherein the dielectric constant estimation section re-estimates the estimated dielectric constant based on the buried object information output by the buried object identification section, the aperture synthesis processing section receives an aperture synthesis processing re-execution flag from outside the buried object discrimination device, and determines whether to re-execute the aperture synthesis processing based on the estimated dielectric constant re-estimated by the dielectric constant estimation section based on the aperture synthesis processing re-execution flag, and if re-execution is decided, the buried object identification section re-executes the buried object identification processing based on the aperture synthesis exploration data and the underground exploration data obtained by the aperture synthesis processing re-execution by the aperture synthesis processing section, and the depth conversion section converts the position of the buried object on the time axis to a depth of the buried object based on the buried object information obtained by re-executing the buried object identification processing and the estimated dielectric constant re-estimated by the dielectric constant estimation section, and outputs the buried object information with depth updated to the converted depth.

4. The buried object discrimination device according to claim 3, further comprising a selection unit having a function of outputting an aperture synthesis processing re-execution flag, a buried object identification re-execution flag, and a depth conversion re-execution flag for controlling re-execution of the aperture synthesis processing unit, the buried object identification unit, and the depth conversion unit based on the underground exploration data, the estimated relative dielectric constant, and the aperture synthetic exploration data, wherein the selection unit comprises: a theoretical underground exploration data generation unit for calculating theoretical underground exploration data based on the estimated relative dielectric constant; a match rate calculation unit for comparing the underground exploration data with the theoretical underground exploration data and calculating a match rate therebetween; a convergence rate calculation unit for calculating a convergence rate of the aperture synthetic exploration data; a depth conversion re-execution flag generation unit for holding the estimated relative dielectric constant and the match rate, and generating the depth conversion re-execution flag; an aperture synthesis processing re-execution flag generation unit for holding the match rate and generating the aperture synthesis processing re-execution flag; and a buried object identification re-execution flag generation unit for holding the convergence rate and generating the buried object discrimination re-execution flag.

5. The buried object discrimination device according to claim 4, wherein said selection section outputs at least one of said coincidence rate and said convergence rate to an outside of said buried object discrimination device.

6. The buried object discrimination device according to claim 1, characterized in that the buried object discrimination device has a buried object learning model generation unit, the buried object learning model generation unit comprising: buried object presence information including information that at least the buried object is present or not present in the exploration area from which the underground exploration data was acquired; a learning data generation unit which generates learning data based on the underground exploration data and the aperture synthetic exploration data; and a learning device which trains a buried object learning model based on a group of accumulated learning data; and the buried object identification unit identifies the buried object using the buried object learning model.

7. The buried object discrimination device according to claim 1, characterized in that the buried object discrimination device has a buried object learning model generation unit, which is equipped with a learning data generation unit that constantly stores the underground exploration data and the aperture synthetic exploration data in an exploration data storage unit, inputs buried object presence information including the known locations of the buried objects, generates learning data from the exploration data group stored in the exploration data storage unit obtained based on the buried object presence information and the buried object presence information and stores the learning data in the learning data storage unit, and a learning device that trains a buried object learning model based on the accumulated learning data group, and the buried object identification unit identifies the buried objects using the buried object learning model.

8. A buried object discrimination device as claimed in claim 1, further comprising a data compression section which receives the input of the underground exploration data, compresses the underground exploration data and outputs it to the relative dielectric constant estimation section, the aperture synthesis processing section and the buried object identification section, respectively, thereby reducing the calculation costs of the relative dielectric constant estimation section, the aperture synthesis processing section and the buried object identification section.

9. A buried object discrimination device as described in claim 1, further comprising a resizing processing unit that receives input of the underground exploration data, resizes the underground exploration data and the aperture synthetic exploration data output from the aperture synthesis processing unit to a predetermined size to generate resized underground exploration data and resized aperture synthetic exploration data, and outputs the resized underground exploration data to the buried object identification unit, thereby allowing the same buried object identification unit to be used regardless of the size of the underground exploration data.

10. The buried object discrimination device according to claim 9, further comprising a data division unit which receives input of the underground exploration data, divides the underground exploration data to generate divided underground exploration data, and outputs the divided underground exploration data to the dielectric constant estimation unit, the aperture synthesis processing unit, and the buried object identification unit, respectively, thereby making it possible to reduce the calculation costs of the dielectric constant estimation unit, the aperture synthesis processing unit, the resizing processing unit, and the buried object identification unit for the divided underground exploration data without reducing the amount of information contained in the underground exploration data.

11. A buried object discrimination device as described in claim 1, wherein at least one of the dielectric constant estimation unit and the aperture synthesis processing unit determines whether the underground exploration data includes antenna structure information including the antenna structure of the underground radar device that generated the underground exploration data, and if the antenna structure information is present, selects the radar reflection formula corresponding to the antenna structure from radar reflection formulas previously defined for each antenna structure, and estimates the dielectric constant based on the radar reflection formula or performs the aperture synthesis processing on the underground exploration data.

12. The buried object discrimination device according to claim 11, wherein the buried object discrimination device has a buried object learning model generation unit, the buried object learning model generation unit constantly accumulates the underground exploration data and the aperture synthetic exploration data, and is equipped with a learning data generation unit which generates learning data based on buried object presence information including information that the buried object is at least present or absent in the exploration area from which the underground exploration data was acquired and on the accumulated exploration data group, and a learning device which extracts a learning data group relating to the same type of antenna structure based on the antenna structure from the accumulated learning data, and learns a buried object learning model for each antenna structure based on the learning data group, wherein the buried object identification unit identifies the buried object by utilizing the buried object learning model relating to the antenna structure of the input underground exploration data from among the accumulated buried object learning models.

13. A buried object management system comprising the buried object discrimination device of claim 1 and a buried object visualization system which visualizes the buried object information, the buried object visualization system comprising: a buried object information storage device which stores the buried object information output from the buried object identification section of the buried object discrimination device; a map information storage device which holds map information; and an information display device which integrates and displays some or all of the buried object information stored in the buried object information storage device and the map information stored in the map information storage device based on the position of the buried object information.

14. A buried object discrimination method executed by a buried object discrimination device which discriminates buried objects underground based on underground exploration data obtained by transmitting and receiving electromagnetic waves underground, comprising: a first step of inputting the underground exploration data provided from an external source; a second step of estimating the relative dielectric constant of the underground medium based on the input underground exploration data; a third step of performing aperture synthesis processing on the underground exploration data based on the estimated relative dielectric constant; a fourth step of performing an identification process to identify the buried object based on the underground exploration data and the aperture synthesis exploration data obtained by performing the aperture synthesis processing on the underground exploration data; and a fifth step of outputting information about the buried object obtained by the identification process as buried object information.

Images