Estimation result display system

Abstract

To display an estimation result intuitively and in an easy-to-understand way.SOLUTION: A top view creation unit 51 extracts, for each of rectangular areas obtained by dividing a horizontal plane into rectangles of a predetermined size, a rectangular parallelepiped area having the highest presence probability estimated by an estimation system, from among a plurality of rectangular parallelepiped areas present immediately below the rectangular areas, based on the position in a depth direction of the extracted rectangular parallelepiped area, sets the hue of pixels corresponding to the rectangular areas in a top view image to a hue corresponding to the position in the depth direction, and based on the presence probability estimated by the estimation system for the extracted rectangular parallelepiped area, sets the brightness of the pixels to brightness corresponding to the presence probability.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] The present invention relates to an estimation result display system that displays the results of estimating the locations of underground pipes such as water pipes, gas pipes, and electric conduits. [Background technology]

[0002] Patent Document 1 describes a radar-type underground exploration device that uses the reflection of pulse waves to explore buried objects underground. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Publication No. 11-190779 Summary of the Invention [Problem to be solved by the invention]

[0004] Conventionally, information obtained by a probe device such as that described in Patent Document 1 has required a skilled engineer to visually analyze waveform images, which requires time and effort for the analysis work. Therefore, it is possible to analyze this using artificial intelligence such as convolutional neural networks (CNN). The results of the analysis using artificial intelligence are expressed, for example, as the probability that a buried pipe exists at a certain location. It is also possible to explore the underground in three dimensions by moving the antenna of the exploration device in two dimensions instead of one, or by providing multiple antennas lined up in a row on the exploration device and moving the antennas in one dimension in a direction different from the direction in which they are lined up. However, when analyzing the results of a three-dimensional underground search, for example, the underground is divided into rectangular parallelepipeds of a given size (for example, cubes with sides of 10 cm), and the probability of the existence of buried pipes is obtained for each area. Because the amount of information is so large, it is difficult to display these results in a way that is intuitively understandable. The present invention aims to solve such problems, for example. [Means for solving the problem]

[0005] The estimation result display system displays estimation results estimated by an estimation system that estimates the probability of a buried pipe being present in each of rectangular parallelepiped regions obtained by dividing the underground into rectangular parallelepiped regions of a predetermined size, and includes a top view creation unit that creates a top view image of the underground from above, and a top view display unit that displays the top view image created by the top view creation unit. For each rectangular region obtained by dividing a horizontal plane into rectangles of a predetermined size, the top view creation unit extracts the rectangular parallelepiped region that has the highest existence probability estimated by the estimation system from among multiple rectangular parallelepiped regions that exist directly below the rectangular parallelepiped region, sets the hue of pixels corresponding to the rectangular region in the top view image to a hue corresponding to a position in the depth direction based on the position in the depth direction of the extracted rectangular parallelepiped region, and sets the brightness of the pixels to a brightness corresponding to the existence probability based on the existence probability estimated by the estimation system for the extracted rectangular parallelepiped region. The estimation result display system may further include a buried pipe display unit that displays a line representing the buried pipe by superimposing it on the top view image displayed by the top view display unit, and a buried pipe position input unit that inputs a position where the buried pipe is located in the top view image displayed by the top view display unit. The buried pipe display unit may move the line to the position input by the buried pipe position input unit and display it. The estimation result display system may further include a buried pipe depth estimation unit that estimates the depth position of the buried pipe based on the position input by the buried pipe input unit. The estimation result display system includes a cross-sectional view creation unit that creates a cross-sectional view image representing a cross section obtained by cutting the underground with a vertical plane, and a cross-sectional view display unit that displays the cross-sectional view image created by the cross-sectional view creation unit. The buried pipe display unit may further display a mark representing the buried pipe by superimposing it on the cross-sectional image displayed by the cross-sectional view display unit. The buried pipe position input unit may further input a position where the buried pipe is located in the cross-sectional image displayed by the cross-sectional view display unit. The buried pipe display unit may display the mark at the position input by the buried pipe position input unit. [Effects of the Invention]

[0006] According to the estimation result display system, the probability of the existence of a buried pipe is expressed by the brightness of the pixel, and the depth of the buried pipe is expressed by the hue of the pixel, thereby improving the visibility of the information at a glance and improving the visibility of the estimation results. The user can correct the position of the buried pipe in the estimated results, thereby improving the accuracy of the estimation. [Brief explanation of the drawings]

[0007] [Figure 1] FIG. 1 is a schematic diagram showing an example of a buried object detection system. [Figure 2] FIG. 1 is a block diagram showing an example of an estimation system and an estimation result display system. [Figure 3] FIG. 10 is a diagram showing an example of a top view image. [Figure 4] FIG. 10 is a flowchart showing an example of an estimation result display process. DETAILED DESCRIPTION OF THE INVENTION

[0008] 1 collects information on the location of buried objects by transmitting radio waves underground and receiving the radio waves reflected by the buried objects. The buried object exploration system 10 includes, for example, a cart 11, multiple radar exploration devices 21, and an exploration position measurement device 31. The cart 11 has, for example, wheels 12 and can move on the ground 80. The cart 11 may be moved by human power, may be towed by a car or the like, or may be self-propelled.

[0009] The radar exploration device 21 is mounted on the cart 11 and moves together with the cart 11. The radar exploration device 21 has, for example, an antenna 22, and emits pulsed radio waves into the ground and receives the radio waves reflected by underground buried objects such as buried objects 81 and buried pipes 82. The distance to the underground buried object can be determined from the time taken from emission to reception of the radio waves. The antenna 22 may be divided into a transmitting antenna and a receiving antenna. The multiple radar exploration devices 21 are arranged, for example, in a line in a direction different from the direction in which the carriage 11 moves (preferably a direction perpendicular to the direction in which the carriage 11 moves). The multiple radar exploration devices 21 are configured to not interfere with each other even when simultaneously exploring, for example, by transmitting and receiving radio waves at different frequencies. This allows the data for one line to be acquired simultaneously. The buried object detection system 10 may have one radar detection device 21 that can move in a direction different from the direction in which the cart 11 moves, rather than having multiple radar detection devices 21. In this case, by moving the radar detection device 21 at a speed that is relatively fast compared to the moving speed of the cart 11, it is possible to acquire one row of data in the same way as when multiple radar detection devices 21 are included. Alternatively, the buried object detection system 10 may have a plurality of radar detection devices 21 that are movable in a direction different from the direction in which the carriage 11 moves.

[0010] The search position measuring device 31 is mounted on the bogie 11 and moves together with the bogie 11. The search position measuring device 31 measures the position of the radar search device 21, which moves as the bogie 11 moves. The search position measuring device 31 may, for example, calculate the travel distance of the bogie 11 from the number of rotations of the wheels 12, or may calculate the travel direction and speed of the bogie 11 based on acceleration measured by an acceleration sensor or the like. Alternatively, the position of the bogie 11 may be measured using a positioning device such as a Global Positioning System (GPS) receiver. When the radar search device 21 moves relative to the bogie 11, the search position measuring device 31 may calculate the absolute position of the radar search device 21 from the absolute position of the bogie 11 and the relative position of the radar search device 21 with respect to the bogie 11.

[0011] 2 estimates the position of buried pipes buried underground based on information collected by the buried object exploration system 10. The estimation system 40 is, for example, a computer, and estimates the position of the buried pipes using artificial intelligence such as a convolutional neural network by having a processing unit execute a program stored in a storage device. For example, the estimation system 40 divides the underground search area into rectangular regions (rectangular regions) of a predetermined size (for example, a cube with each side measuring 10 cm), and calculates the probability that a buried pipe exists within each rectangular region. Furthermore, for example, when a region where there is a high probability that a buried pipe exists is linearly continuous, the estimation system 40 estimates that a buried pipe exists over that range.

[0012] 2 displays the estimation results estimated by the estimation system 40. The estimation result display system 50 is, for example, a computer, and realizes the functional blocks described below by having a processing unit execute a program stored in a storage device. The estimation result display system 50 includes, for example, a top view creation unit 51, a top view display unit 52, a cross-sectional position input unit 53, a cross-sectional view creation unit 54, a cross-sectional view display unit 55, a buried pipe display unit 56, a buried pipe position input unit 57, a buried pipe depth estimation unit 58, and a meaning input unit 59.

[0013] The top view creation unit 51 creates a top view image based on the estimation results obtained by the estimation system 40. As shown in FIG. 3, the top view image represents the search range as viewed from directly above. The top view image is, for example, a bitmap image, and has a plurality of pixels arranged vertically and horizontally in a checkerboard pattern. Each pixel corresponds to a rectangular area obtained by dividing a horizontal plane into rectangles of a predetermined size (for example, squares with sides of 10 cm). Here, the brightness of a pixel represents the probability that a buried pipe exists directly below that rectangular area. For example, a brightness of 100% represents a 100% probability of existence, and a brightness of 0% represents a 0% probability of existence. The hue of a pixel represents the depth of the buried pipe. For example, if the hue of a pixel is 0 degrees (red), it indicates that the buried pipe is 0 cm deep; if the hue is 64 degrees (green), it indicates that the buried pipe is 0.8 m deep; and if the hue is 128 degrees (blue-purple), it indicates that the buried pipe is 1.6 m deep. For example, for each of the rectangular areas described above, the top view creation unit 51 extracts multiple rectangular parallelepiped areas directly below the rectangular area based on the estimation results obtained by the estimation system 40. Then, from among the multiple extracted rectangular parallelepiped areas, the rectangular parallelepiped area with the highest probability of the existence of a buried pipe estimated by the estimation system 40 is selected. The top view creation unit 51 sets the brightness of the corresponding pixel based on the probability of the existence of a buried pipe in the selected rectangular parallelepiped area. In addition, the top view creation unit 51 sets the hue of the corresponding pixel based on the depth position of the selected rectangular parallelepiped area.

[0014] The top view display unit 52 displays the top view image created by the top view creation unit 51 on a display device such as a liquid crystal display.

[0015] The cross section position input unit 53 uses an input device such as a mouse to input the position where the user wants to display a vertical cross section in the top view image displayed by the top view creation unit 51 .

[0016] The cross-sectional view creation unit 54 creates a vertical cross-sectional view image based on the position input by the cross-sectional position input unit 53. The vertical cross-sectional view image represents a cross-section cut in the search range by a vertical plane passing through the position input by the cross-sectional position input unit 53, as viewed from the side. The vertical plane may be a vertical plane parallel to a predetermined direction among the vertical planes passing through the position input by the cross-sectional position input unit 53, or a vertical plane parallel to any direction selected by the user. The cross-sectional view creation unit 54 may also create multiple vertical cross-sectional views based on multiple different vertical planes. Based on the estimation results obtained by the estimation system 40, the cross-sectional view creation unit 54 extracts, for example, from among the above-mentioned multiple rectangular parallelepiped areas, multiple rectangular parallelepiped areas that intersect with a vertical plane determined based on the positions input by the cross-sectional position input unit 53, and sets the brightness of pixels in the vertical cross-sectional view image based on the probability of the existence of buried pipes in the extracted rectangular parallelepiped areas.

[0017] The cross-sectional view display unit 55 displays the vertical cross-sectional view image created by the cross-sectional view creation unit 54 on a display device such as a liquid crystal display.

[0018] The buried pipe display unit 56 displays a line indicating the buried pipe, superimposed on the top view image displayed by the top view display unit 52, based on the estimation result estimated by the estimation system 40. The buried pipe display unit 56 also displays a mark indicating the buried pipe, superimposed on the cross-sectional view image displayed by the cross-sectional view display unit 55, based on the estimation result estimated by the estimation system 40.

[0019] The buried pipe position input unit 57 inputs the actual position of the buried pipe specified by the user using an input device such as a mouse when it is known, for example, from the location of an excavation or manhole that a buried pipe exists in a position different from the position where the buried pipe display unit 56 displays a line or mark. The buried pipe position input unit 57 inputs the horizontal position of the actual buried pipe, for example, by the user dragging and dropping the line displayed by the buried pipe display unit 56 overlaid on the top view image displayed by the top view display unit 52. The buried pipe position input unit 57 also inputs the actual position of the buried pipe within the cross section, for example, by the user moving or deleting the mark displayed by the buried pipe display unit 56 overlaid on the cross section image displayed by the cross section display unit 55, or by adding a new mark. The buried pipe display unit 56 moves the lines superimposed on the top view image displayed by the top view display unit 52 and the marks superimposed on the cross-sectional view image displayed by the cross-sectional view display unit 55 to the positions input by the buried pipe position input unit 57. The buried pipe position input by the buried pipe position input unit 57 may be fed back to the estimation system 40, for example, so that the artificial intelligence used by the estimation system 40 for estimation can learn the position.

[0020] The buried pipe depth estimation unit 58 recalculates the depth position of the buried pipe based on the buried pipe position input by the buried pipe position input unit 57. For example, for multiple rectangular regions corresponding to multiple pixels where the line displayed by the buried pipe display unit 56 overlaps with the top view image displayed by the top view display unit 52, multiple rectangular parallelepiped regions directly below each rectangular region are acquired, and the depth positions of all the acquired rectangular parallelepiped regions are weighted by the probability of the existence of a buried pipe estimated by the estimation system 40 for each rectangular parallelepiped region and averaged to estimate the depth position of the buried pipe at the corrected position. Alternatively, taking into account the possibility that the buried pipe is inclined, the depth positions of the rectangular parallelepiped regions may be weighted by the probability of the existence of a buried pipe and linearly approximated by the least squares method. The buried pipe display unit 56 moves the mark superimposed on the cross-sectional view image displayed by the cross-sectional view display unit 55 to the position estimated by the buried pipe depth estimation unit 58.

[0021] The meaning input unit 59 inputs meanings (annotations) for the estimation results estimated by the estimation system 40. For example, if it is clear from the positions of excavations and manholes that no buried pipe exists at the location where the buried pipe display unit 56 displays a line or mark, the user inputs that this is a false detection by checking a "false detection" checkbox using an input device such as a mouse. Conversely, if it is clear that a buried pipe actually exists at the location where the buried pipe display unit 56 displays a line or mark, the user inputs that this is the correct detection by checking a "correct answer" checkbox using an input device such as a mouse. The meaning input by the meaning input unit 59 may be fed back to the estimation system 40, for example, so that the artificial intelligence used by the estimation system 40 for estimation can learn the meaning.

[0022] Next, the estimation result display process will be described. As shown in Figure 4, the estimation result display process includes, for example, a dataset display step S11, an operation input step S12, an upload selection determination step S21, an upload execution step S22, an inference selection determination step S31, an inference execution step S32, a dataset selection determination step S41, a detailed display step S42, a top view selection determination step S51, a cross-sectional view update step S52, a correction selection determination step S61, a correction execution step S62, a download selection determination step S71, a download execution step S72, a meaning-assignment selection determination step S81, a meaning-assignment execution step S82, an additional learning selection determination step S91, and an additional learning execution step S92.

[0023] First, in a data set display step S11, the estimation result display system 50 displays a list of registered data sets on the display screen of a display device such as a liquid crystal display, etc. Then, the process proceeds to an operation input step S12. In the operation input step S12, the inference result display system 50 waits for an operation from the user, and when the operation from the user is input, the process proceeds to the upload selection determination step S21.

[0024] In the upload selection determination step S21, the inference result display system 50 determines whether the user's operation input in the operation input step S12 is to select an upload process, such as clicking an upload button. If the user's operation is to select an upload process, the system proceeds to the upload execution step S22. Otherwise, the system proceeds to the inference selection determination step S31. In the upload execution step S22, the estimation result display system 50 executes the upload process of the dataset, and then returns to the dataset display step S11.

[0025] In the inference selection determination step S31, the inference result display system 50 determines whether the user's operation input in the operation input step S12 is to select an inference process, such as clicking an inference start button. If the user's operation is to select an inference process, the system proceeds to the inference execution step S32. Otherwise, the system proceeds to the dataset selection determination step S41. In the inference execution step S32, the estimation system 40 estimates the position of the buried pipe based on the registered data set, and then the process returns to the data set display step S11.

[0026] In the dataset selection determination step S41, the estimation result display system 50 determines whether the user's operation input in the operation input step S12 is to select one dataset from the list of datasets displayed in the operation input step S12. If the user's operation is to select a dataset, that dataset is selected and the process proceeds to the detail display step S42. Otherwise, the system determines whether a dataset has already been selected. If a dataset has already been selected, the process also proceeds to the detail display step S42. If no dataset has been selected, the process returns to the dataset display step S11. In the detail display step S42, the estimation result display system 50 displays details of the selected data set. For example, the top view image generated by the top view creation unit 51 is displayed by the top view display unit 52, the cross-sectional view generated by the cross-sectional view creation unit 54 is displayed by the cross-sectional view display unit 55, and lines and marks representing the buried pipe are displayed superimposed on these by the buried pipe display unit 56. Then, the process proceeds to the top view selection determination step S51.

[0027] In the top view selection determination step S51, the cross-section position input unit 53 determines whether the user's operation input in the operation input step S12 is to select a point on the top view, such as by clicking on any point on the plan view. If the user's operation is to select a point on the top view, the process proceeds to the cross-section view update step S52. Otherwise, the process proceeds to the correction selection determination step S61. In the cross-sectional view update step S52, the cross-sectional view creation unit 54 updates the cross-sectional view displayed in the detail display step S42 based on the point selected by the user's operation input in the operation input step S12. For example, a cross-sectional view cut by a plane including the point selected in the top view selection determination step S51 is displayed. Then, the process returns to the data set display step S11.

[0028] In the correction selection determination step S61, the estimation result display system 50 determines whether the user's operation input in the operation input step S12 is to select a correction process to correct the position of the buried pipe 82, such as by clicking a pipe position correction button. If the user's operation is to select a correction process, the system proceeds to the correction execution step S62. Otherwise, the system proceeds to the download selection determination step S71. In the correction execution step S62, the buried pipe position input unit 57 executes a correction process. For example, the user inputs an operation to correct the position of the buried pipe, such as dragging and dropping a line representing the buried pipe displayed on the top view. The buried pipe position input unit 57 corrects the estimated horizontal position of the buried pipe 82 in accordance with the input operation. The buried pipe depth estimation unit 58 recalculates the depth of the buried pipe 82 in accordance with the correction of the horizontal position of the buried pipe 82. The buried pipe display unit 56 moves the lines and marks representing the buried pipe that are displayed superimposed on the top view image and cross-sectional view image. Then, the process returns to the data set display step S11.

[0029] In the download selection determination step S71, the inference result display system 50 determines whether the user's operation input in the operation input step S12 is to select a data set download process, such as clicking a download button. If the user's operation is to select a download process, the process proceeds to the download execution step S72. Otherwise, the process proceeds to the semantic selection determination step S81. In the download execution step S72, the estimation result display system 50 executes the download process for the selected dataset, and then returns to the dataset display step S11.

[0030] In the meaning selection determination step S81, the meaning input unit 59 determines whether the user's operation input in the operation input step S12 is to select a meaning process, such as clicking an annotation button. If the user's operation is to select a meaning (annotation) process, the process proceeds to the meaning execution step S82. Otherwise, the process returns to the data set display step S11. In the meaning assignment execution step S82, the meaning assignment input unit 59 performs a meaning assignment process on the selected data set. For example, the user inputs whether or not the result of the estimation system 40 using artificial intelligence to determine the location of a buried pipe is correct. The estimation result display system 50 adds the input meaning to the data set and stores it. Then, the process proceeds to the additional learning selection determination step S91.

[0031] In the additional learning selection determination step S91, the meaning input unit 59 prompts the user to select whether or not to have the artificial intelligence undergo additional learning, based on the meaning input in the meaning execution step S82. If the user selects additional learning, the process proceeds to the additional learning execution step S92. If the user does not select additional learning, the process returns to the dataset display step S11. In the additional learning execution step S92, the estimation system 40 causes the artificial intelligence to perform additional learning based on the meaning input in the meaning assignment execution step S82. This improves the judgment accuracy of the artificial intelligence. Thereafter, the process returns to the data set display step S11. Instead of allowing the user to choose whether or not to perform additional learning on the artificial intelligence, the additional learning execution step S92 may always be executed, or a determination may be made as to whether or not a predetermined condition is met, and if the condition is met, the additional learning execution step S92 may be executed.

[0032] As described above, when checking the results of the assessment in a top view, the brightness represents the probability of the existence of buried pipes, and the hue represents the depth of the buried pipes. This prevents the screen from becoming cluttered with numbers, as occurs when simply displaying each numerical value as text, or when the results overlap, preventing the numbers from overlapping and making the screen less legible. Furthermore, if the estimation results obtained by the estimation system are incorrect, the pipe positions in the top view and cross-sectional view states can be corrected from an expert's perspective, thereby improving the accuracy of pipe position estimation.

[0033] The above-described embodiment is an example for facilitating understanding of the present invention. The present invention is not limited thereto, and includes various modifications, changes, additions, or omissions without departing from the scope defined by the appended claims. This can be easily understood by those skilled in the art from the above description. [Explanation of symbols]

[0034] 10 Buried object detection system, 11 Cart, 12 Wheel, 21 Radar detection device, 22 Antenna, 31 Detection position measurement device, 40 Estimation system, 50 Estimation result display system, 51 Top view creation unit, 52 Top view display unit, 53 Cross section position input unit, 54 Cross section creation unit, 55 Cross section display unit, 56 Buried pipe display unit, 57 Buried pipe position input unit, 58 Buried pipe depth estimation unit, 59 Meaning input unit, 80 Ground, 81 Buried object, 82 Buried pipe, S11 Data set display process, S12 Operation input process, S21 Upload selection determination process, S22 Upload execution process, S31 Inference selection determination process, S32 Inference execution process, S41 Data set selection determination process, S42 Detail display process, S51 Top view selection determination process, S52 Cross section update process, S61 A correction selection determination step, S62 a correction execution step, S71 a download selection determination step, S72 a download execution step, S81 a meaning assignment selection determination step, S82 a meaning assignment execution step, S91 an additional learning selection determination step, S92 an additional learning execution step.

Claims

1. An estimation result display system displays the estimation results of an estimation system that estimates the probability that buried pipes exist in each rectangular parallelepiped area obtained by dividing the underground into rectangular parallelepiped areas of a predetermined size, a top view creating unit that creates a top view image of the underground; a top view display unit that displays the top view image created by the top view creation unit; Equipped with The top view creation unit divides a horizontal plane into rectangles of a predetermined size, and for each of the rectangular regions, extracting the rectangular parallelepiped region having the highest existence probability estimated by the estimation system from among the plurality of rectangular parallelepiped regions present directly below the rectangular region; setting a hue of a pixel corresponding to the rectangular region in the top view image to a hue corresponding to a position in the depth direction based on a position in the depth direction of the extracted rectangular parallelepiped region; setting the brightness of the pixel to a brightness corresponding to the existence probability based on the existence probability estimated by the estimation system for the extracted rectangular parallelepiped region; Estimation result display system.

2. a buried pipe display unit that displays a line representing the buried pipe by superimposing it on the top view image displayed by the top view display unit; a buried pipe position input unit for inputting the position where the buried pipe is located in the top view image displayed by the top view display unit; Further provided with the buried pipe display unit moves the line to the position input by the buried pipe position input unit and displays it. The estimation result display system according to claim 1.

3. a buried pipe depth estimation unit that estimates the position of the buried pipe in the depth direction based on the position input by the buried pipe position input unit; The estimation result display system of claim 2 further comprising:

4. a cross-sectional view creation unit that creates a cross-sectional view image representing a cross section obtained by cutting the underground along a vertical plane; a cross-sectional view display unit that displays the cross-sectional view image created by the cross-sectional view creation unit; Further provided with The buried pipe display unit further displays a mark representing the buried pipe superimposed on the cross-sectional view image displayed by the cross-sectional view display unit, The buried pipe position input unit further inputs a position where the buried pipe is present in the cross-sectional view image displayed by the cross-sectional view display unit, the buried pipe display unit displays the mark at the position input by the buried pipe position input unit; The estimation result display system according to claim 2 or 3.

Images