3D model generation system

Abstract

To easily and accurately generate a three-dimensional model in a work chamber and to easily and accurately calculate a numerical aperture in a pneumatic caisson method.SOLUTION: The three-dimensional model generating system comprises: a projection device 11 that is attached to the top surface of a work chamber M and that projects a code image 10 in which coordinates of the attaching position is written vertically downward; an imaging device 12 that images the inside of the work chamber M including the code image 10 projected by the projection device 11; and an analysis device 13 that image-processes the image data imaged by the imaging device 12, and generates a three-dimensional model in the work chamber M. A doming phenomenon in the image-processing can be prevented with the code image 10 as a feature point. The three-dimensional model adjusted in scale can be generated by reading the coordinates written in the code image 10.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to a three-dimensional model generation system in a working chamber in the pneumatic caisson method.

Background Art

[0002] The pneumatic caisson method is a method in which an airtight working chamber is provided at the bottom of the caisson, and compressed air corresponding to the interstitial water pressure of the ground is sent into the chamber to prevent the intrusion of groundwater. While excavating and discharging earth and sand in a dry state, the caisson constructed on the ground is gradually sunk. In the pneumatic caisson method, the mechanization of excavation and discharge has advanced, and the entry of workers into the working chamber under high pressure is limited to only very limited operations such as inspection and maintenance of excavation machinery, and unmanned construction or mechanized construction is being promoted, and many achievements have been made so far.

[0003] When sinking the caisson, excavation of the earth and sand on the bottom surface of the working chamber and discharging the earth outside the casing are carried out to reduce the supporting force at the blade edge, and sinking excavation is performed to sink the caisson by the self-weight of the caisson or the like. As the caisson sinks, the remaining excavated soil at the peripheral part of the working chamber bulges, and the supporting force at the blade edge increases. When this supporting force or the like is almost balanced with the self-weight of the caisson or the like, the sinking of the caisson stops.

[0004] In the sinking excavation of the caisson, in order to correctly sink the caisson without inclining or over-sinking it, the supporting force at the blade edge is managed using an index called the opening ratio (A - A') / A, which is the ratio of the area of the opening A - A' obtained by subtracting the area A' of the remaining excavated soil at the peripheral part of the working chamber from the bottom area A of the caisson.

[0005] In calculating the opening ratio, it is important to accurately measure the area of the remaining excavated soil. Conventionally, a caisson worker enters the working chamber, measures the width of the remaining excavated soil with a scale or the like, and calculates the opening ratio manually on the ground or using spreadsheet software incorporated in an electronic computer.

[0006] However, in recent years, technologies have been developed to automatically measure the amount of unexcavated soil and to automatically generate 3D models of the work chamber, in order to prevent health damage caused by caisson workers entering high-pressure work chambers and to monitor the excavation status inside the work chamber in real time.

[0007] For example, Patent Document 1 below discloses a caisson equipped with a distance measuring device such as a laser, which is placed at a reference point on the lower surface side of the caisson's base slab and measures the distance from the slope of the remaining excavated portion of the soil on the lower side near the caisson's cutting edge to the reference point.

[0008] Furthermore, Patent Document 2 discloses a measurement system for measuring the surface shape of unexcavated soil in real time, comprising: a rotation angle measuring instrument that measures the rotation angle of the rotating body of an excavator in real time and outputs the measured value; and a distance / depression angle measuring instrument that is installed on the rotating body of an excavator and measures the distance to the incident point of the laser beam and the depression angle of the irradiation direction of the laser beam in real time by deflecting the laser beam vertically while irradiating the laser beam toward the periphery of the work chamber, and outputs the measured value. [Prior art documents] [Patent Documents]

[0009] [Patent Document 1] Japanese Patent Publication No. 2015-229826 [Patent Document 2] Japanese Patent Publication No. 2020-90798 [Overview of the project] [Problems that the invention aims to solve]

[0010] However, in the systems described in Patent Documents 1 and 2 above, since a laser is used as the distance measuring device, all objects irradiated with laser light are converted into point clouds, and the process of removing objects unnecessary for calculating the aperture ratio (e.g., temporary soil, heavy machinery, running rails, etc.) is time-consuming and laborious. Furthermore, if there are puddles of water on the excavation surface inside the caisson, there are problems such as the difficulty in measuring distance due to the reflection characteristics of the laser light.

[0011] Furthermore, in Patent Document 1, a laser-emitting distance measuring device is configured to travel along rails installed on the ceiling of the workroom. However, since the ceiling of this workroom is equipped with and constantly operating caisson shovels, earth buckets, etc., there is a drawback in that changing the setup between excavation and laser measurement becomes extremely complicated.

[0012] Therefore, the main objective of the present invention is to enable the simple and accurate generation of a 3D model of the working chamber in the pneumatic caisson construction method, as well as to enable the simple and accurate generation of the opening ratio. Certainly The objective is to provide a system for generating 3D models that can be calculated. [Means for solving the problem]

[0013] To solve the aforementioned problems, the present invention according to claim 1 is a system for generating a three-dimensional model of the working chamber in a pneumatic caisson construction method, The aforementioned It is installed on the ceiling of the workroom. In addition, multiple caissons are provided at intervals along the direction in which the caisson's cutting edge extends, within the XY plane of the workroom ceiling. Mounting position XY The code image with the coordinates written on it is oriented vertically downwards. Project onto the ground surface A projection device for displaying images, A height measuring device that measures the vertical distance from the projection device to the projected code image in order to obtain the Z coordinate, They are attached to the ceiling surface of the aforementioned workroom and are provided in multiples at intervals along the direction in which the caisson's cutting edge extends. An imaging device that captures images of the workroom, including the code image projected by the projection device, The image captured by the aforementioned imaging device Analyzing a large number of image data, each Image data The 3D coordinates and imaging direction are estimated, and the code images present in each image data are detected. The code images common to adjacent image data are used as feature points to perform matching of the image data, and point cloud generation is performed by multi-view image measurement. A 3D model generation system is provided, characterized by comprising an analysis device that generates a 3D model of the workroom.

[0014] In the invention described in claim 1 above, a three-dimensional model of the workroom is generated by image processing of the image data captured by the imaging device using the analysis device, and the aperture ratio is determined from this three-dimensional model.

[0015] In image processing, if image data has few feature points, a doming phenomenon occurs where distortion increases towards the edges, potentially leading to failure in generating a 3D model. Therefore, in this invention, clear feature points are added to the image data captured by the imaging device by including a code image projected by the projection device. This prevents the doming phenomenon in image processing, enabling the easy and accurate generation of a 3D model.

[0016] Furthermore, since the image data captured by the imaging device is dimensionless, it is necessary to adjust it to the actual size in order to determine the aperture ratio. Therefore, in the present invention, the code image is The XY of the mounting position of the projection device in the XY plane of the workroom ceiling. By writing coordinates and then reading those written coordinates, it becomes possible to quickly, easily, and accurately generate a 3D model with accurate scaling.

[0017] Thus, in this invention, by capturing image data of the region including the code image projected by the projection device using the imaging device, feature points are added to each image data, the doming phenomenon when a 3D model is generated can be suppressed, and the code image is written to it. XY coordinate and Z coordinate from the height measuring device By reading this data, the scale of the 3D model can be automatically adjusted. Furthermore, by detecting the code images present in each image data, matching the image data using the code images common to adjacent image data as feature points, and generating a point cloud through multi-view image measurement, a 3D model of the workroom is generated. This technology allows for the accurate and easy calculation of the opening ratio from a 3D model of the working chamber in the pneumatic caisson construction method.

[0018] Claim 2 As part of the present invention, a 3D model generation system according to claim 1 is provided, wherein the imaging device is provided in which a plurality of rows of imaging devices are provided at intervals along the direction in which the caisson's cutting edge extends, and the rows of imaging devices are provided in a direction perpendicular to the direction in which the caisson's cutting edge extends.

[0019] In the above claim 2 In the described invention, an imaging device array arranged along the direction in which the blade edge of the caisson extends is provided in multiple rows in a direction orthogonal to the direction in which the blade edge of the caisson extends. Therefore, the situation of the remaining soil at the periphery of the working chamber can be grasped more accurately.

[0020] Claim 3 Regarding the present invention according to this claim, the projection device projects a , randomly placed patterns feature image as onto an area other than the code image, which can be a feature point during image processing, and there is provided a three-dimensional model generation system according to claim 1.

[0021] The invention described in claim 3 above adds randomly arranged patterns as feature images, in addition to using code images as feature points, when mapping image data together. As described above, when generating a three-dimensional model by the analysis device, feature points are required for the image data captured to suppress the occurrence of the doming phenomenon. Book In the invention, in addition to the above-described code image, more feature points are added by projecting a feature image by the projection device. The feature image is not particularly limited as long as it is a characteristic image, but a randomly arranged pattern is preferable, and for example, a checkerboard pattern, a vine pattern, an animal pattern, etc. can be used.

Effect of the Invention

[0022] As described in detail above, according to the present invention, a three-dimensional model in the working chamber in the pneumatic caisson method can be generated simply and accurately, and the aperture ratio can be calculated simply and accurately.

Brief Description of the Drawings

[0023] [Figure 1] It is a longitudinal sectional view showing the lower part of the caisson 1 in the pneumatic caisson method to which the three-dimensional model generation system according to the present invention is applied. [Figure 2] It is a plan view of the working chamber M for explaining the aperture ratio. [Figure 3] It is a plan view showing the arrangement of equipment on the ceiling of the working chamber. [Figure 4] This is a partial perspective view showing the periphery of workroom M. [Figure 5] This diagram shows the image acquisition area of ​​the image data captured by the imaging device 12. [Figure 6] This is a perspective view showing an example of generating a 3D model (mesh model). [Figure 7] This is a perspective view showing an example of generating a combined 3D model (mesh model). [Figure 8] This is a projection image of code image 10 with feature image 14 added. [Modes for carrying out the invention]

[0024] Embodiments of the present invention will be described in detail below with reference to the drawings.

[0025] As shown in Figure 1, the pneumatic caisson method involves creating a working chamber M, an airtight space enclosed by the bottom plate 1a of the caisson 1 and the cutting edges 1b that hang down around it. Compressed air is supplied into this working chamber M through an air supply pipe 3 embedded in the side wall 1c of the caisson 1, creating an atmospheric pressure that matches the pore water pressure of the ground, thereby preventing groundwater seepage. Excavation of the ground G and removal of soil are carried out within the working chamber M, while simultaneously constructing caissons to connect to the upper side on the ground, gradually sinking the caisson 1 to a predetermined depth.

[0026] Furthermore, a hollow material shaft 4 is erected vertically from the base slab 1a of the caisson 1 for the removal of excavated soil, and a man shaft 6 for communication between workers is provided adjacent to it.

[0027] For excavating the ground G within the workroom M, a remotely controlled excavator, specifically a shovel 9 that can move along a travel rail 8 fixed to the ceiling surface (underside of the base slab 1a) of the workroom M, is installed, and unmanned construction is introduced. The soil excavated by the shovel 9 is loaded into an earth bucket (not shown) that is lowered through the material shaft 4 by an automatic loading machine (not shown) and transported outside.

[0028] When the aforementioned shovel 9 is remotely operated to excavate the ground G, the bearing force acting on the inner surface of the cutting edge 1b decreases due to the remaining soil 2 at the periphery of the working chamber M in the ground G. As a result, the caisson 1 settles due to its own weight (including the weight of load water injected into the inside of the caisson 1). As the caisson 1 settles, the remaining soil 2 rises, increasing the bearing force on the cutting edge 1b. When this bearing force is roughly balanced with the caisson 1's own weight, the caisson 1 stops settling.

[0029] As shown in Figure 2, the ratio of the opening area A-A', which is obtained by subtracting the area A' of the remaining excavated soil 2 at the periphery of the work chamber M from the bottom area A of caisson 1, is defined as the opening ratio (A-A') / A, and the settlement excavation of caisson 1 is managed using this opening ratio.

[0030] This invention is a system for generating a three-dimensional model of the working chamber M in the pneumatic caisson construction method. By measuring the area A' of the remaining soil 2 from the three-dimensional model of the remaining soil 2 generated by this system, the opening ratio can be easily and accurately calculated.

[0031] As shown in Figures 1 and 3, the 3D model generation system according to the present invention consists of a projection device 11 attached to the ceiling surface (lower surface of the base plate 1a) of the workroom M, which projects a code image 10 (see Figure 4) on which the coordinates of its position are written, in a vertical downward direction; an imaging device 12 that images the inside of the workroom M, including the code image 10 projected by the projection device 11; and an analysis device 13 that takes in the image data captured by the imaging device 12, processes the image, and generates a 3D model of the workroom M. Furthermore, an illumination device (not shown) is appropriately provided to illuminate the inside of the workroom M.

[0032] In image processing, if image data has few feature points, the locations of the vertices of the mesh model shift, and a doming phenomenon occurs where distortion increases towards the edges, potentially leading to failure in generating a 3D model. Therefore, in this invention, by capturing image data with the imaging device 12 so as to include the code image 10 projected into the working chamber M, clear feature points are provided in the image data, preventing the doming phenomenon in image processing. In the pneumatic caisson method, the condition of the soil inside the working chamber M changes constantly depending on the soil type at the depth of excavation. For example, at depths where soft ground is excavated, soil tends to accumulate on the wall surface of the working chamber M (inner surface of the cutting edge 1b), and the wall surface of the working chamber M may be hidden by soil. For this reason, if characteristic markers are fixedly placed on the wall surface of the working chamber M, the piled-up soil may become an obstacle, or imaging may not be possible due to the effects of mud splashes, etc. Therefore, in this invention, the code image 10, which will be the feature points, is projected onto the ground G using the projection device 11. This makes it possible to add characteristic points regardless of the soil conditions and without causing contamination.

[0033] Since the image data captured by the imaging device 12 is dimensionless, it is necessary to adjust it to actual size in order to determine the aperture ratio from the area of ​​the remaining excavated soil 2. In this invention, since coordinates are written on the code image 10 projected into the workroom M, a well-scaled 3D model can be quickly, easily, and accurately generated by reading these coordinates.

[0034] As described above, in the present invention, since image data including the code image 10 projected by the projection device 11 is captured by the imaging device 12, clear feature points are added to each image data, the doming phenomenon when a 3D model is generated can be suppressed, and the scale of the 3D model can be automatically adjusted by reading the coordinates written in the code image 10. Therefore, the opening ratio can be calculated accurately and easily from the 3D model of the work chamber M in the pneumatic caisson construction method.

[0035] Next, the components of the 3D model generation system according to the present invention will be described in more detail. The projection device 11 is a so-called projector that projects an image by shining light onto a projection surface. The projection device 11 is fixed at a predetermined xy coordinate position with respect to the ceiling surface of the workroom M (the lower surface of the base plate 1a of the caisson 1) as the xy plane, and projects the code image 10 vertically downward onto the upper surface (ground) of the ground G from there. One code image 10 is projected by one projection device 11, and the center of the code image 10 coincides with the xy coordinate of the projection device 11.

[0036] Multiple projection devices 11 are provided at intervals along the direction in which the cutting edge 1b of the caisson 1 extends. As a result, multiple code images 10 are provided at intervals along the direction in which the cutting edge 1b extends, on the ground G in the workroom M, at a predetermined distance from the cutting edge 1d at the lower end of the cutting edge 1b. The spacing of the projection devices 11 in relation to the direction in which the cutting edge 1b extends is arbitrary, but it is preferable to arrange them so that at least one code image 10 is included in each image data captured by the imaging device 12. Multiple projection devices 11 may be arranged in rows perpendicular to the direction in which the cutting edge 1b extends.

[0037] The code image 10 is not particularly limited as long as the written data is readable by a dedicated reader, but two-dimensional codes such as QR codes (registered trademark) or barcodes are used. The coordinates written in the code image 10 are the coordinates (x, y) where the projection device 11 is installed in the plane (xy plane) of the ceiling of the workroom, as shown in Figure 3.

[0038] On the other hand, the height coordinate (z-coordinate) needs to be measured each time because the distance between the top surface of the ground G and the ceiling surface of the workroom M changes depending on the soil type and other factors. For this reason, in the present invention, a height measuring device (not shown) for measuring the vertical distance from the projection device 11 to the code image 10 is provided in or near the projection device 11. As the height measuring device, a string-like object with a scale attached is suspended from the ceiling surface and read by analyzing the image data captured by the imaging device 12, or a laser distance meter, ultrasonic distance meter, etc., can be used. This height measuring device may be built into the projection device 11, or it may be placed separately in the vicinity of the projection device 11.

[0039] The image data captured by the imaging device 12 may be a series of still images or a video. The imaging device 12 uses a digital camera and is equipped with a transmission means capable of transmitting the captured image data to the ground-based analysis device 13. As the imaging device 12, for example, a 360-degree camera can be used that can be remotely controlled automatically or manually to change the camera's orientation in the horizontal (pan) and vertical (tilt) directions. The 360-degree camera is a camera whose orientation can be changed over at least almost the entire circumference (350 to 360°) in the horizontal direction, and can be changed approximately 90° downwards in the vertical direction, with the horizontal direction being 0°.

[0040] The imaging device 12 is mounted on the ceiling surface of the workroom M. Preferably, the imaging device 12 is mounted so that a magnet is equipped on its fixing surface, allowing it to be fixed to the steel plate ceiling of the workroom M by the magnetic force of the magnet. This allows for easy installation on the ceiling surface and easy fine adjustment of its position. By fixing the imaging device 12 to the ceiling surface of the workroom M, imaging by the imaging device 12 does not affect the excavation work of the shovel 9 moving along the rail 8, thereby improving work efficiency.

[0041] As shown in Figure 3, multiple imaging devices 12 are provided at intervals along the direction in which the cutting edge 1b of the caisson 1 extends. This allows for accurate assessment of the condition of the remaining excavated soil 2 at the periphery of the work chamber M.

[0042] In the example configuration shown in Figure 3, the imaging device 12 consists of multiple rows of imaging devices spaced apart along the direction in which the caisson 1's cutting edge 1b extends, and multiple rows of imaging devices arranged perpendicular to the direction in which the caisson 1's cutting edge 1b extends. By providing multiple rows of imaging devices, the condition of the remaining excavated soil 2 around the edge of the work chamber can be grasped more accurately, and it is also possible to respond to changes in the on-site environment, such as the presence of obstacles or inability to take pictures due to backlighting. In the example configuration shown in Figure 3, the imaging device rows are provided in two rows, but three or more rows may be provided.

[0043] The spacing between adjacent imaging devices 12 is arbitrary, but is preferably 1 to 5 m, and more preferably 2 to 3 m. If this spacing is too small, the number of imaging devices 12 required will increase, resulting in an excessive amount of image data, which will increase the time and cost of analysis. On the other hand, if this spacing is too large, the overlap between image data captured by adjacent imaging devices 12 will be small, which may reduce the accuracy of the analysis.

[0044] Figure 5 shows the image data acquisition range D1 by the first row of imaging devices closer to the blade opening 1b, and the image data acquisition range D2 by the second row of imaging devices further from the blade opening 1b. The imaging devices 12 arranged in the first row capture image data such that more than half, preferably more than 60%, of the area of ​​adjacent imaging devices 12 overlap. The imaging devices 12 arranged in the second row encompass the entire range D1 of the image data of the first row and capture image data over a wider area.

[0045] Of the image data captured by the imaging device 12, it is preferable to exclude image data containing obstacles or other objects unnecessary for image analysis from the image analysis, thereby enabling an accurate understanding of the excavation status within the workroom M. Furthermore, while obstacles often appear in the area to be photographed at the work site, since the imaging device 12 is permanently installed, if the obstacle is temporary, it can be dealt with by acquiring images after the obstacle has passed. Even if the obstacle remains for a while, it is preferable to exclude the obstacle as much as possible when generating the 3D model of the workroom M by covering the data loss through parallax interpolation of adjacent imaging devices 12.

[0046] Image data captured by the imaging device 12 is transmitted to an analysis device 13, such as a computer, installed in a ground control room, where image processing is performed. This image processing uses the principle of aerial triangulation (AT) to generate a 3D mesh model based on a large amount of image data. By using known Structure from Motion (SfM) software for image processing, it is possible to automatically generate the 3D model.

[0047] In the aforementioned image processing (SfM processing), a large number of image data are analyzed to estimate the 3D coordinates and imaging direction (imaging orientation) of each image data, and feature points (such as the code image 10) present in each image data are detected. By using the feature points that are common to adjacent image data, image data is matched, and a point cloud is generated by multi-view image measurement, thereby generating a 3D model (3D mesh model) for each partitioned area, as shown in Figure 6, for example. Surface textures may be added to the generated 3D mesh model from the image data to make the surface appearance match that of the real object.

[0048] The image processing described above is performed to generate a continuous, unified 3D model across the entire workroom M if the caisson is relatively small, but if the caisson is relatively large, the workroom M is divided into multiple areas, a 3D model is generated for each area, and these 3D models are then joined based on the coordinates of the code image 10. In the example shown in Figure 7, the workroom M is divided into three areas M1, M2, and M3, a 3D mesh model is generated for each area, and then these are joined together.

[0049] To increase the number of feature points during image processing, prevent the doming phenomenon, and improve image processing accuracy, it is desirable to project a feature image 14, which can serve as a feature point during image processing, onto an area other than the code image 10, as shown in Figure 8. The feature image 14 is not particularly limited as long as it is a distinctive image, but randomly arranged patterns are preferred. For example, mottled patterns, arabesque patterns, and animal patterns can be used, as shown in the illustrated example. In addition, randomly arranged shapes such as circles, squares, and stars, or letters and symbols may also be used.

[0050] To perform caisson settlement excavation using the 3D model generation system configured as described above, image data is acquired by the imaging device 12, a 3D model of the work chamber M is generated by the analysis device 13, the area of ​​the remaining excavated soil 2 is determined from this 3D model, and the opening ratio is calculated. The calculated current opening ratio is compared with the design value, and based on this, the next excavation plan is made and the excavation instruction is given to the shovel 9. Once the predetermined excavation is completed, the procedure of acquiring image data again by the imaging device 12 is repeated. [Explanation of Symbols]

[0051] 1...Caisson, 2...Excavated soil, 3...Air supply pipe, 4...Material shaft, 6...Manshaft, 8...Travel rail, 9...Shovel, 10...Code image, 11...Projection device, 12...Imaging device, 13...Analysis device

Claims

1. A system for generating a three-dimensional model of the working chamber in the pneumatic caisson construction method, A projection device is installed on the ceiling surface of the aforementioned workroom, and multiple such devices are provided at intervals along the direction in which the caisson's cutting edge extends. The device projects a code image containing the XY coordinates of the installation position in the XY plane of the workroom ceiling vertically downwards and onto the ground. A height measuring device that measures the vertical distance from the projection device to the projected code image in order to obtain the Z coordinate, An imaging device is installed on the ceiling surface of the aforementioned workroom and is provided in multiple locations at intervals along the direction in which the caisson's cutting edge extends, and captures images of the workroom, including the code image projected by the projection device. A three-dimensional model generation system characterized by comprising an analysis device that analyzes a large number of image data captured by the aforementioned imaging device to estimate the three-dimensional coordinates and imaging direction of each image data, detects the code images present in each image data, matches the image data using the code images common to adjacent image data as feature points, and generates a point cloud by multi-view image measurement to generate a three-dimensional model of the workroom.

2. The 3D model generation system according to claim 1, wherein the imaging device is provided in multiple rows of imaging devices spaced apart along the direction in which the caisson's cutting edge extends, and these rows are arranged in a direction perpendicular to the direction in which the caisson's cutting edge extends.

3. The 3D model generation system according to claim 1, wherein the projection device projects randomly arranged patterns, which can serve as feature points during image processing, as feature images onto an area other than the code image.

Images