Buried object detection device, and buried object detection method

Abstract

To provide a buried object detection device capable of detecting buried objects more accurately over a wider area, and a buried object detection method.SOLUTION: The buried object detection device includes: a radar data processing unit that converts multiple radar data measured by an underground measuring device into images: a buried object detection unit that detects buried objects from the images of the radar data; an underground feature detection unit that detects a characteristic portion that is different from the buried object in the images of the radar data; an alignment unit that calculates alignment information for multiple radar data alignment based on the detection results from the underground feature detection unit; and a result integration unit that integrates the buried object detection results based on the output result of the buried object detection unit and the output result of the alignment unit.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to an embedded object detection device and an embedded object detection method.

Background Art

[0002] There is a need for subsurface surveys due to the aging of underground infrastructure. It is possible to provide customers with a service for creating an underground map that visualizes the subsurface structure measured with radar data or the like.

[0003] The buried object linear extraction device of Patent Document 1 extracts a starting point for tracking the linear shape of a buried object based on data obtained by three-dimensionally representing the reflected wave intensity of electromagnetic waves, and performs a process of tracking the linear shape of the buried object from the starting point. A selection process is performed to exclude the tracked linear shape of the buried object based on a predetermined criterion, and a simplification process is performed to group adjacent tracking results among the selected plurality of tracking results as one group.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] In the process of tracking the linear shape of the buried object in Patent Document 1, since the tracking of the buried object is performed multiple times on the three-dimensional data, there is no problem of misalignment between the plurality of tracking results. However, since multiple trackings are performed on the same sensing data, when detecting a buried object using data sensed at different positions or timings, the merit of improving the detection performance of the buried object cannot be obtained.

[0006] Therefore, there is a problem of providing an embedded object detection device and an embedded object detection method that can more accurately detect a wide range of embedded objects.

Means for Solving the Problems

[0007] One of the typical aspects of the present invention for solving the above problems is the following buried object detection device. This buried object detection device includes a radar data processing unit, a buried object detection unit, a subsurface feature detection unit, a registration unit, and a result integration unit. The radar data processing unit images a plurality of radar data measured by a subsurface measurement device. The buried object detection unit detects buried objects from the image of the radar data. The subsurface feature detection unit detects a feature portion different from the buried object in the image of the radar data. The registration unit calculates registration information regarding the registration of a plurality of radar data based on the detection result from the subsurface feature detection unit. The result integration unit integrates the detection results of the buried objects based on the output result of the buried object detection unit and the output result of the registration unit.

[0008] One of the typical aspects of the present invention for solving the above problems is the following buried object detection method. This method is performed by a processor executing a program stored in a memory. This method includes a step in which the processor performs an imaging process on a plurality of radar data measured by a subsurface measurement device, a step in which the processor detects a buried object and a feature portion different from the buried object from the imaged radar data, a step in which the processor calculates registration information, which is information regarding the registration of a plurality of radar data, based on the detection result of the feature portion, and a step in which the processor integrates the detection results of the buried objects using the registration information.

Advantages of the Invention

[0009] According to the present invention, by integrating the buried object detection results at a plurality of locations with high precision, buried objects over a wide range can be detected more accurately. Note that problems, configurations, and effects other than those described above will be clarified by the description of the following embodiments.

Brief Description of the Drawings

[0010]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6A

Figure 6B

Figure 7

Figure 8

Figure 9

Figure 10A

Figure 10B

Figure 11

Mode for Carrying Out the Invention

[0011] Hereinafter, embodiments of the present invention will be described with reference to the drawings. The embodiments are examples for explaining the present invention, and for the sake of clarity of explanation, appropriate omissions and simplifications have been made. The present invention can be implemented in various other forms. Unless otherwise particularly limited, each component may be singular or plural. The positions, sizes, shapes, ranges, etc. of the components shown in the drawings may not represent the actual positions, sizes, shapes, ranges, etc. in order to facilitate the understanding of the invention. For this reason, the present invention is not necessarily limited to the positions, sizes, shapes, ranges, etc. disclosed in the drawings. As examples of various types of information, they may be described using expressions such as "table", "list", "queue", etc., but the various types of information may be represented by data structures other than these. For example, various types of information such as "XX table", "XX list", "XX queue", etc. may be referred to as "XX information". When explaining identification information, expressions such as "identification information", "identifier", "name", "ID", "number", etc. are used, and these can be mutually substituted. When there are a plurality of components having the same or similar functions, they may be described by attaching different subscripts to the same reference numeral. Also, when it is not necessary to distinguish these plurality of components, the subscripts may be omitted in the description. In the embodiments, the processing performed by executing a program may be described. Here, the computer executes the program by a processor (e.g., CPU, GPU), and performs the processing defined by the program while using storage resources (e.g., memory) and interface devices (e.g., communication ports), etc. Therefore, the entity of the processing performed by executing the program may be the processor. Similarly, the entity of the processing performed by executing the program may be a controller, device, system, computer, or node having a processor. The entity of the processing performed by executing the program only needs to be an arithmetic unit, and may include a dedicated circuit for performing a specific process. Here, the dedicated circuit is, for example, an FPGA (Field Programmable Gate Array), an ASIC (Application Specific Integrated Circuit), a CPLD (Complex Programmable Logic Device), etc. The program may be installed in a computer from a program source. The program source may be, for example, a program distribution server or a computer-readable storage medium. When the program source is a program distribution server, the program distribution server includes a processor and a storage resource for storing the program to be distributed, and the processor of the program distribution server may distribute the program to be distributed to other computers. Also, in an embodiment, two or more programs may be realized as one program, or one program may be realized as two or more programs.

[0012] <First Embodiment> FIG. 1 is a schematic diagram showing an example of an embedded object detection device (hereinafter referred to as a detection device). The detection device 1a in the first embodiment integrates the detection results of radar data obtained by multiple scans and displays the results in a wide range to the user. The detection device 1a, for example, performs imaging processing on a plurality of radar data, performs detection of embedded objects and underground feature detection, performs alignment using the results of the underground feature detection, and integrates the detection results of a plurality of embedded objects based on the alignment information obtained as a result of the alignment.

[0013] The vehicle 2 continuously measures while scanning the underground space by the measuring device 21. The measuring device 21 is provided with a ground penetrating radar 22. The ground penetrating radar 22 measures the underground space (that is, the ground) by irradiating in the underground direction. The measurement data of the underground space of the ground penetrating radar is sent to the detection device 1a.

[0014] The detection device 1a includes, for example, a radar data processing unit 10, an embedded object detection unit 11, an underground feature detection unit 12, an alignment unit 13, a result integration unit 14, and a display unit 15. First, the radar data processing unit 10 processes a plurality of radar data sent from the vehicle 2 and forms an image. The embedded object detection unit 11 and the underground feature detection unit 12 perform embedded object detection and underground feature detection on each of the plurality of images sent from the radar data processing unit 10. Then, the embedded object detection unit 11 sends the embedded object detection result to the result integration unit 14. The detection result of the underground feature of the underground feature detection unit 12 is sent to the alignment unit 13.

[0015] Further, the alignment unit 13 performs alignment of a plurality of radar data based on the detection result of the subsurface feature detection unit 12. The alignment information that is the output of the alignment unit 13 is sent to the result integration unit 14.

[0016] The result integration unit 14 integrates the buried object detection results in the images of the plurality of radar data based on the output of the buried object detection unit 11 and the output of the alignment unit 13. The integrated result is sent to the display unit 15. The display unit 15 causes the integrated result to be displayed on a display device.

[0017] FIG. 2 is a diagram showing an example of the hardware configuration of the detection device. The detection device 1a includes, for example, a storage device 41, a CPU (Central Processing Unit) 42, a memory 43, and a communication device 44.

[0018] In the storage device 41, for example, a radar data processing program 100, a buried object detection program 101, a subsurface feature detection program 102, an alignment program 103, and a result integration program 104 are stored. Note that the programs may be recorded on a storage medium MM such as a USB (Universal Serial Bus) memory.

[0019] The CPU 42 realizes each function by reading each program (100 to 104) from the storage device 41 into the memory 43 and executing it. The memory 43 is a volatile storage medium such as a RAM (Random Access Memory), for example. The communication device 44 is connected to be capable of two-way communication with the vehicle 2, the database 6, and the user terminal 5 including the display device 51 via the network 45. Further, the display device 105 is a type of output device and outputs the calculation result and the like in the detection device 1a. For this reason, the display device 105 can be realized by a monitor, a touch panel, or the like. However, the display device 105 can be omitted. Furthermore, the detection device 1a may have an input device such as a keyboard or a mouse for the user to operate. Note that the display device 105 and the input device may be integrally realized like a touch panel.

[0020] The user terminal 5 is a portable terminal such as a smartphone. Note that the user terminal 5 may be a personally-owned computer equipped with a monitor of the display device 51. The display device 51 is not limited to being provided in the user terminal 51, and may be provided in the detection device 1a.

[0021] Note that the data transfer between the processes of each program (100 to 104) may be performed by the CPU 42 via the memory 43 or the storage device 41 or the like. The detection device 1a is not limited to being wirelessly communicably connected to the display device 51 and the database 6 via the network 45, and may be wired communicably connected.

[0022] The measurement data of the measurement device 21 may be stored in the database 6. The detection device 1a may execute each function (10 to 15) by appropriately acquiring data such as the measurement data stored in the database 6.

[0023] The data transfer from the vehicle 2 to the database 6 may be sequentially processed via the network 45, for example. The database 6 may have a configuration included in a database server or a configuration included in the detection device 1a. The detection device 1a and the display device 51 may be connected via a display interface such as an RGB (Red Green Blue) cable, for example.

[0024] FIG. 3 is a flowchart of the processing of the detection device 1a in the first embodiment. Specifically, various processes are mainly performed by the processor (in the embodiment, the CPU 42). The buried object detection unit 11 and the subsurface feature detection unit 12 perform imaging processing on the plurality of radar data obtained by scanning a plurality of times (S1). Hereinafter, the measurement data obtained in the process (S1) will be described with reference to FIG. 4.

[0025] In FIG. 4, the x-axis direction indicates the direction in which the vehicle moves. The y-axis direction indicates the width direction of the vehicle. The z-axis direction indicates the depth direction of the subsurface.

[0026] First, Run 1, Run 2, and Run N in the figure are different driving lines although they are in close proximity to each other. Due to driving errors, there may be a slight deviation in the y-axis positions of the start and end points of each run. Also, each driving line does not have to be completely parallel. The radar data for each run is stored in the order of the measured time. The radar data processing converts the ground cross-section into a grayscale image according to the amplitude of the reflected wave of the radar.

[0027] Returning to FIG. 3, the description will continue. The buried object detection unit 11 individually detects buried objects for each of the plurality of radar data output by the radar data processing unit 10 (S2_1). The details of the buried object detection process in step S2_1 will be described with reference to FIG. 5.

[0028] FIG. 5 is a diagram for explaining the buried object detection process in step S2_1. The processor individually inputs the radar data from Run 1 to Run N into the buried object detection model. The buried object detection model can be, for example, a buried object detection model generated by learning data with the positions of buried objects labeled as teacher data. The type and position information of the buried object can be estimated by the buried object detection model.

[0029] Regarding the position information, it may be the bounding box of the object or a point representing the object, such as the vertex of a hyperbola representing the buried object, the center point of a circle approximating the buried object, etc. Each radar data outputs detection results 1 to N of the buried object by the buried object detection model. Similarly, the buried object detection model outputs features 1 to N extracted from the buried object regions detected from each radar data. These extracted features 1 to N are represented by, for example, a heatmap indicating the likelihood of the detected buried object. In this case, higher-brightness features are extracted from regions that are more likely to be buried objects.

[0030] Returning to FIG. 3, the description will continue. The subsurface feature detection unit 12 detects subsurface features for the radar data obtained by multiple scans (S2_2). The subsurface feature detection process in step S2_2 will be described with reference to FIG. 6.

[0031] FIG. 6 is an explanatory diagram of the subsurface feature detection process. In the subsurface feature detection process, subsurface features are detected. FIG. 6A is an explanatory diagram of the subsurface features. As shown in FIG. 6A, the subsurface features are features in the radar data and relate to feature portions other than the detection targets of the buried object detection unit 11. For example, at locations where there are boundaries between different paving layers (i.e., where there are continuous geological changes), such as the paving layer of a road, there are obvious reactions in the radar data. Also, this boundary reaction can be observed similarly at nearby driving positions. Such boundary portions are cited as an example of the subsurface features.

[0032] In addition to the boundary lines of geological changes, other subsurface features can include cavities, specific regions (e.g., regions with buried pipes having a circular cross-section in the image), regions with large local luminance changes in the radar data image (e.g., regions with many gaps in the image and large luminance changes), and any other buried objects that can be detected on the radar data image.

[0033] FIG. 6B is a diagram for explaining the subsurface feature detection process in step S2_2. The processor inputs the data measured in each run from run 1 to run N in order into the subsurface feature detection model. The subsurface feature detection model can be, for example, a subsurface feature detection model generated by learning data with labeled positions such as the boundaries of geological strata with different geology as teacher data. The position information of the subsurface features can be estimated by the subsurface feature detection model. Each radar data outputs detection results 1 to N of the subsurface features by the subsurface feature detection model.

[0034] Returning to FIG. 3, the description will be continued. The alignment unit 13 inputs the detection results of the underground features of a plurality of radar data and calculates alignment information between adjacent radar data. Regarding the alignment process in step S3, specifically, when there are the detection result S of the underground features at the S-th time and the detection result S-1 of the underground features at the (S-1)-th time (1 < S < N + 1), the processor matches the area of the underground features in the detection result S at the S-th time with the area of the underground features in the detection result S-1 at the (S-1)-th time and calculates an alignment matrix. In addition, in the matching of the underground features extracted from the plurality of radar data in step S3, by using auxiliary information such as GPS and the self-position estimation information of the traveling vehicle together, the matching accuracy can be improved and a more accurate alignment matrix can be calculated.

[0035] Next, the result integration unit 14 integrates the results based on the detection result of the buried object detection unit 11 and the output result of the alignment unit 13 regarding the alignment information (S4). The result integration process in step S4 will be described with reference to FIG. 7.

[0036] FIG. 7 is a diagram showing an example of the flow of the result integration process in step S4 in the first embodiment. In step S41, the processor acquires the result of the buried object detection unit 11 and the result of the alignment unit 13 (the results at the S-th time and the (S-1)-th time (1 < S < N + 1)). Also, in step S42, the processor determines whether alignment information such as an alignment matrix can be acquired. If Yes, the process transitions to step S43 to perform processing using the alignment information. If No, the process transitions to step S44.

[0037] Also, in step S43, the processor performs alignment of the detected buried objects using the alignment information. For example, the processor aligns the data at the (S - 1)-th time with the data at the S-th time as a reference.

[0038] When the alignment information cannot be acquired, in step S44, the processor performs a matching process based on the detection result of the buried object detection unit 11. The details will be described with reference to FIG. 8.

[0039] FIG. 8 is a diagram showing an example of the flow of the buried object detection result matching process in step S44. In step S441, the processor acquires the buried object detection result of the buried object detection unit 11 and the characteristics of the buried object. Also, in step S442, it is determined whether each detection result is included in the matching area. The matching area is an area for searching whether a plurality of buried objects detected from radar data obtained by different runs match.

[0040] The matching area will be described with reference to FIG. 9. FIG. 9 is a diagram for explaining the buried object detection unit result matching. In FIG. 9, an example of acquiring radar data by three runs is given for explanation. For example, in the first run, for the buried objects (31, 32, 33), the detection areas are D31, D32, D33. Here, the matching area is set larger than the detection area of the buried object. For example, it can be an area with a size 200% of the rectangular area including the detection area. Matching areas (M31, M32, M33) are similarly generated for each of D31, D32, D33 where a buried object is detected.

[0041] Returning to FIG. 8 to continue the explanation. In S442, if two buried objects to be compared are included in the matching area (Yes), the process transitions to S443. If not included in the matching area, the process transitions to step S444.

[0042] For example, in the first and second data in FIG. 9, since the buried objects (31, 31_2) are each included in the matching area M31, the processor performs processing related to the buried objects (31, 31_2) in S443. Since the buried objects (32, 34) are not each included in the matching area M32, the processor performs processing related to the buried objects (32, 34) in S444.

[0043] In step S443, the processor determines whether the features of the two embedded objects match. For example, assuming that the two embedded objects included in the matching region are represented by feature vectors of the same dimensionality, the Mahalanobis distance between the two features is calculated, and if the distance is less than or equal to a preset threshold, it is determined that they match. If the features match (Yes), the process transitions to step S445. If the features do not match (No), the process transitions to step S444.

[0044] For example, in the first and second data in FIG. 9, since it is determined that the feature amounts of the embedded objects (31, 31_2) match, the processor performs processing related to the embedded objects (31, 31_2) in S445. Also, although included in the same matching region M33, since it is determined that the feature amounts of the embedded objects (33, 35) do not match, the processor performs processing related to the embedded objects (33, 35) in S444.

[0045] Also, for example, in the first and third data in FIG. 9, the embedded objects (32, 32_3) are respectively included in the matching region M32, and since it is determined that the feature amounts of the embedded objects (32, 32_3) match, the processor performs processing related to the embedded objects (32, 32_3) in S445.

[0046] In step S445, the processor performs processing to match the detection results of the embedded objects in the radar data during different travels. The embedded objects to be matched are assigned the IDs of the already detected embedded objects. For the embedded objects with these IDs, the matching region and the features of the embedded objects are updated to the matching region and the features of the matched embedded objects. Or, without updating the features of the embedded objects, the two matched features can be respectively retained.

[0047] For example, in the first and second data in FIG. 9, the processor performs a process of matching the detection results of the embedded objects (31, 31_2), and assigns an ID indicating that the embedded object has already been detected. Then, in the third data, the processor can update the features of the matching area (M31_3) and the embedded object (31_3) to the features of the matching area (M31_2) and the embedded object (31_2) of the matched embedded object. Or, the features of the embedded object (31_3) can be kept without being updated while maintaining the matching features.

[0048] That is, for example, in the second time, the detection area (D31_2) of the embedded object (31_2) is in the first matching area M31. Further, since the features of the detection area (D31_2) and the features of the detection area D31 match, the same ID is assigned to the embedded object (31_2) and the embedded object 31, and the feature area and the matching area for the embedded object 31 are updated to D31 and M31. Here, the feature area corresponds to the detection area and is an area having features. The feature area of the third embedded object (31_3) is in the second matching area (M31_2). Further, since the features of the detection area (D31_3) and the detection area (D31_2) match, the ID of the embedded object 31 is assigned to the embedded object (31_3).

[0049] In step S444, when two embedded objects detected from different driving data do not match, the processor newly generates a matching area of the embedded object and adds it to the list of embedded objects.

[0050] A specific example will be described with reference to FIG. 9. Since the embedded object 34 detected in the second time is not in the matching area M32, the processor creates a new ID and a matching area M34 for the embedded object 34. Similarly, although the embedded object 35 detected in the second time is in the matching area M33, since the features of the detection area D35 and the features of the detection area D33 do not match, the processor creates a new ID and a matching area M35 for the embedded object 35.

[0051] Returning to FIG. 8, the description will be continued. In step S446, the processor outputs the result of the matching. In step S45 of FIG. 7, the processor integrates and outputs the detection result of the buried object. In S5 of FIG. 3, the display unit 15 outputs the result of the result integration unit 14. That is, in step S5, the processor outputs and presents the result integrated by the result integration unit 14. Here, an example of the display content by the display unit 15 will be described with reference to FIG. 10.

[0052] FIG. 10 shows an example of the display mode by the display unit, and this display includes information indicating the buried object.

[0053] As shown in FIG. 10A, the detection device 1a integrates the detection results of the radar data obtained by multiple runs, and the display device 51 displays the detection results of the entire area. FIG. 10A shows, as an example, the case of integrating the results of detecting buried objects with radar data obtained by three runs. In this example, the detection device 1a performs buried object detection processing on each radar data, and integrates each running area by alignment or the like, thereby integrating and displaying the information of the buried objects in the wide area detected in each of these three runs.

[0054] Note that the detection device 1a may generate data for displaying so as to be able to identify each buried object. In other words, the detection device 1a may generate data for displaying in a manner of combining data having the same ID. Therefore, as an example, the processor may generate data for surrounding and displaying the same buried object with a line (dotted line in FIG. 10). Note that the display mode may be changed as appropriate.

[0055] Also, as shown in FIG. 10B, for each run, the integrated result of the detected buried objects can be updated and displayed. The display device 51, for example, displays the buried objects (31, 32, 33) detected in the first run. Then, the buried object (31_2) that matched in the second run and the buried object (31_1) in the first run are combined and displayed. Further, the new buried objects (34, 35) in the second run are additionally displayed. Then, the buried object (31_3) that can be matched in the third run and the buried object (31_1) in the first run and the buried object (31_2) in the second run are combined and displayed. Also, the buried object (32_3) and the buried object (32_1) in the first run are combined and displayed.

[0056] By including the display unit 15, the detection device 1a can cause the display device 51 to display the result of detecting buried objects using a plurality of run data. Thereby, the user can more accurately grasp the positional relationship of the structures in the three-dimensional space.

[0057] <Second Embodiment> Next, the second embodiment will be described. Note that the description similar to that of the first embodiment may be omitted. The detection device 1b in the second embodiment performs the detection of buried objects by using, in addition to the processing of the first embodiment, the combined data of a plurality of radar data obtained by a plurality of scans. In the second embodiment, the detection device 1b integrates the detection result of the buried objects with respect to the combined data of the plurality of radar data and the detection result of the buried objects in the first embodiment. FIG. 11 shows an example of the detection device in the second embodiment.

[0058] In addition to the radar data of multiple runs, the radar data synthesis unit 23 of the vehicle 2 also inputs data obtained by synthesizing the data of these multiple runs to the detection device 1b. The synthesized data buried object detection unit (11_1) has a function of detecting buried objects from the synthesized radar data, and the synthesized radar data is input to the synthesized data buried object detection unit (11_1). In the detection of buried objects by the synthesized data buried object detection unit (11_1), the processor can detect the bounding box of the buried object or representative points of the buried object, such as the center point, vertex, etc., using a model that has learned data with the positions of the buried objects labeled in advance as teacher data.

[0059] The detection result from the synthesized data buried object detection unit (11_1) is sent to the result integration unit B (13_1). Note that this result integration unit B (13_1) may be described as a synthesized data result integration unit. In the result integration unit B (13_1), the detection result of the buried object in the first embodiment and the detection result from the synthesized data buried object detection unit (11_1) are integrated. First, integration of the same type of buried objects is performed. The processor integrates buried objects with overlapping detection positions or buried objects whose distance is closer than a preset threshold into one, and outputs the others as separate buried objects. Then, the processor can generate data for displaying each buried object so that it can be identified on the display unit 15.

[0060] Note that, for example, a synthesized data buried object detection program related to the synthesized data buried object detection unit (11_1) and a synthesized data result integration program related to the result integration unit B (13_1) may be stored in the storage device 41. Note that the program may be recorded on a storage medium MM such as a USB (Universal Serial Bus) memory. Also, the synthesized data may be stored in the database 6. The detection device 1b may execute processing by appropriately acquiring data such as the synthesized data stored in the database 6.

[0061] According to the above-described second embodiment, it is expected to improve the detection accuracy of buried objects with synthetic data having improved image quality. For example, when the buried object signal disappears in the synthetic data, the detection accuracy of the buried object can be complementarily improved by combining the detection of the buried object in each scan data.

[0062] According to the first and second embodiments, buried object detection is performed on each data of a plurality of scan data, and each detection result is integrated based on the position of the buried object detected in each scan data and features extracted from the measurement data of the buried object, and the result can be presented.

[0063] In addition, by performing alignment based on the continuity of the radar data values associated with changes in the ground other than the buried objects to be detected (such as the paving layer), the detection results of the buried objects from a plurality of radar data can be integrated with high accuracy without using external data such as GPS or ground photographed images.

[0064] And by using the detection devices (1a, 1b), it is possible to contribute from a social perspective by improving the detection accuracy of buried objects in social infrastructure projects.

[0065] Although the embodiments have been described above, the present invention is not limited to the above-described embodiments, and various modifications and equivalent configurations within the scope of the appended claims are included. For example, the above-described embodiments have been described in detail for easy understanding of the present invention, and the present invention is not necessarily limited to those having all the configurations described. Also, for example, for a part of the configuration of the embodiment, addition, deletion, or replacement of other configurations may be made.

[0066] The detection devices (1a, 1b) may, as an example, acquire radar data based on measurements of the same type of radar and perform processing. On the other hand, the detection devices (1a, 1b) may also acquire radar data based on measurements of radars with different performances (for example, radars with different resolutions, noises, etc.) and perform processing. By appropriately including radar data based on radars with different performances, an improvement in detection accuracy using an embedded object detection model and a subsurface feature detection model can be expected.

[0067] In the embodiment, the CPU 42 is described as the processor, but it may be appropriately changed as long as it can perform appropriate processing, and the processor may be configured using, for example, other semiconductor devices.

[0068] The storage device 41 can be configured as appropriate, and can be configured using, for example, an HDD (Hard Disk Drive), a ROM (Read Only Memory), etc. The storage device 41 can store data used for processing, and may store, for example, an embedded object detection model, a subsurface feature detection model, etc.

[0069] The embedded object detection model may be learned using learning data to detect an embedded object as a detection target (objective), and the subsurface feature detection model may be learned using learning data to detect an embedded object different from the above-mentioned embedded object.

Explanation of Reference Numerals

[0070] 1 ··· Detection device 2 ··· Vehicle 10 ··· Radar data processing unit 11 ··· Embedded object detection unit 12 ··· Subsurface feature detection unit 13 ··· Alignment unit 14 ··· Result integration unit 15 ··· Display unit 21 ··· Measuring device 22 ··· Ground penetrating radar

Claims

1. A radar data processing unit that images a plurality of radar data measured by a subsurface measurement device, An embedded object detection unit that detects an embedded object from an image of the radar data, A subsurface feature detection unit that detects a feature portion different from the embedded object in the image of the radar data, An alignment unit that calculates alignment information regarding alignment of a plurality of radar data based on a detection result from the subsurface feature detection unit, A result integration unit that integrates detection results of the embedded object based on an output result of the embedded object detection unit and an output result of the alignment unit, An embedded object detection device comprising the above.

2. The embedded object detection device according to Claim 1, wherein the subsurface feature detection unit detects a portion of continuous geological change as the feature portion. An embedded object detection device characterized by the above.

3. The embedded object detection device according to Claim 1, wherein the subsurface feature detection unit detects at least one of a cavity, an embedded pipe, a region with many gaps, and an embedded object as the feature portion. An embedded object detection device characterized by the above.

4. The embedded object detection device according to Claim 1, wherein when the alignment information cannot be obtained, the result integration unit integrates the detection results of the embedded object by performing matching of the embedded objects detected by a plurality of radar data by the embedded object detection unit. An embedded object detection device characterized by the above.

5. The embedded object detection device according to Claim 1, further comprising a display unit that displays the integration result of the result integration unit, wherein when the embedded object detection device acquires a plurality of radar data, the detection results of the embedded object are integrated each time the radar data is acquired, and the display unit displays the result of the above integration. An embedded object detection device characterized by the above.

6. The embedded object detection device according to Claim 1, further comprising a synthetic data embedded object detection unit that detects an embedded object from data obtained by synthesizing a plurality of radar data, and a synthetic data result integration unit that integrates the detection result of the synthetic data embedded object detection unit and the integration result of the result integration unit. An embedded object detection device characterized by the above.

7. An embedded object detection method performed by a processor executing a program stored in a memory, wherein the processor performs a step of imaging a plurality of radar data measured by a subsurface measurement device, The step in which the processor detects an embedded object and detects a feature part different from the embedded object from the imaged radar data; The step in which the processor calculates alignment information, which is information regarding the alignment of a plurality of radar data, based on the detection result of the feature part; The step in which the processor integrates the detection result of the embedded object by using the alignment information; An embedded object detection method characterized by including the above.

8. The embedded object detection method according to claim 7, wherein the processor detects a part of continuous geological change as the feature part. An embedded object detection method characterized by the above.

9. The embedded object detection method according to claim 7, wherein the processor detects at least one of a cavity, an embedded pipe, a region with many gaps, and an embedded object as the feature part. An embedded object detection method characterized by the above.

10. The embedded object detection method according to claim 7, in the step of integrating the detection result of the embedded object, when the processor cannot obtain the alignment information, the processor integrates the detection result of the embedded object by performing matching of the embedded objects detected by a plurality of radar data. An embedded object detection method characterized by the above.

11. The embedded object detection method according to claim 7, wherein the embedded object detection method is a method further performed by using a display device used by a user, in the step of integrating the detection result of the embedded object, when acquiring radar data a plurality of times, the processor integrates the detection result of the embedded object each time one radar data is acquired, and the display device further includes the step of displaying the result of the above integration. An embedded object detection method characterized by the above.

12. The embedded object detection method according to claim 7, including the step in which the processor detects an embedded object from data obtained by synthesizing a plurality of radar data, and the step in which the processor further integrates the detection result of the embedded object based on the synthesized data and the result of integrating the detection result of the embedded object by using the alignment information. An embedded object detection method characterized by the above.

Images