Estimation device, program, and estimation method

Abstract

To accurately grasp ground surface shape by simple processing.SOLUTION: An estimation device (100) includes: an acquisition part (101) for acquiring a near-infrared image (Ii) and a visible light image (Iv); a calculation unit (102) for calculating a normalization difference vegetation index of a virtual grid (GK); a generation part (103) for generating a target DSM (3); and a correction part (104) for correcting an elevation value of a third grid (G3) according to a comparison result of the normalization difference vegetation index and a threshold.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to an estimation device, a program, and an estimation method for estimating elevation values in each of a plurality of grids constituting a digital elevation model.

Background Art

[0002] As a typical conventional technique related to topographic survey of land, there is a technique of acquiring three-dimensional point cloud data indicating the topography of the land and gridifying this three-dimensional point cloud data to generate a three-dimensional model. The three-dimensional point cloud data is composed of X and Y coordinate data representing horizontal positions, and Z coordinate data representing elevation.

[0003] When the land targeted by the three-dimensional model includes features, this three-dimensional model becomes a DSM (Digital Surface Model) including the shape of the features. Features are, for example, man-made structures such as buildings and bridges, and vegetation such as trees, which are objects that shield the ground surface. In this case, when trying to grasp the shape of the ground surface, that is, the shape obtained by removing the shape of the features from the topography of the land, a filtering process is performed on the DSM to generate a DEM (Digital Elevation Model). The filtering process is a process of removing measurement data of the feature part from the DSM. The DEM is a three-dimensional model representing the shape of the ground surface.

[0004] By the way, in the generation of the DEM, a high level of proficiency is required for the filtering process, so there is a problem that the generation of the DEM takes more time than that of the DSM. Therefore, in order to solve this problem, technologies for grasping the shape of the ground surface using the DSM have been researched and developed. For example, Patent Document 1 discloses a technique for calculating the elevation value of the ground surface by subtracting the average tree height of a forest area from the elevation values of the forest area using three-dimensional surface model data (DSM).

Prior Art Documents

Patent Documents

[0005] [Patent Document 1] Japanese Patent Application Laid-Open No. 2012-243226 [Summary of the Invention] [Problems to be Solved by the Invention]

[0006] However, when calculating the elevation value of the ground surface using the technique disclosed in Patent Document 1, since it cannot be said that the average tree height in the forest area accurately reflects the vegetation state in the forest area, the accuracy of the obtained elevation value is not sufficient. In addition, since calculating the average tree height requires complex calculations, it takes time for the calculation process of the average tree height. Therefore, the technique disclosed in Patent Document 1 has room for improvement in accurately and effortlessly grasping the shape of the ground surface.

[0007] One aspect of the present invention has been made in view of the above problems, and an object thereof is to enable the shape of the ground surface to be grasped with high accuracy by simpler processing than before. [Means for Solving the Problems]

[0008] In order to solve the above problems, an estimation device according to an aspect of the present invention is an estimation device that estimates the shape of a target ground surface, which is the ground surface to be estimated, and a target vegetation, which is the vegetation on the target ground surface, by virtually dividing each of a plurality of unit spaces obtained by virtually dividing the target ground surface and the target vegetation into a grid shape, and using an arbitrary virtual grid among the plurality of virtual grids as a target grid. The estimation device includes: an acquisition unit that acquires at least an infrared image in which the target ground surface and the target vegetation are imaged; a calculation unit that calculates a correction index indicating the state of vegetation in the virtual grid corresponding to the target grid using the first spectral intensity of a first grid corresponding to the target grid among a plurality of grids obtained by dividing the infrared image into a grid shape; a generation unit that generates a numerical surface model as a three-dimensional model of a surface layer formed by the target ground surface and the surface of the target vegetation using optical information of reflected light reflected by the target ground surface and the target vegetation; and a correction unit that corrects the elevation value of a third grid corresponding to the target grid according to a comparison result between the correction index and a threshold value among a plurality of grids obtained by dividing the numerical surface model into a grid shape.

[0009] According to the above configuration, since the elevation value of the third grid can be corrected only by comparing the correction index with the threshold value, there is no need to separately calculate the average height of the vegetation when correcting the elevation value of the third grid. In addition, since the correction index accurately represents the state of the vegetation, the accuracy of the elevation value after correction by the correction unit is higher than before. From the above, the shape of the ground surface to be estimated can be grasped with high accuracy by a simpler process than before.

[0010] In an estimation device according to an aspect of the present invention, the acquisition unit may acquire a near-infrared image as the infrared image and further acquire the optical information, and the calculation unit may calculate the correction index using the first spectral intensity and a second spectral intensity in a predetermined wavelength region of a second grid corresponding to the target grid included in the optical information.

[0011] According to the above configuration, by calculating the correction index using the spectral intensity of the near-infrared image and the spectral intensity of the light information, the shape of the ground surface to be estimated can be grasped with higher accuracy using simpler processing than conventional methods.

[0012] In the estimation device according to one aspect of the present invention, the calculation unit may calculate a normalized difference vegetation index as the correction index by using a spectral intensity in which the specified wavelength range is red in the visible range as the second spectral intensity, and the correction unit may correct the elevation value according to a comparison result between the normalized difference vegetation index and the threshold value.

[0013] According to the above configuration, the normalized difference vegetation index, which is the most widely used and most reliable of the vegetation indices, is used as the correction index, so the reliability of the elevation value corrected by the correction unit is higher than when other vegetation indices are used as the correction index. This makes it possible to grasp the shape of the ground surface to be estimated with more accuracy using simpler processing than before.

[0014] In the estimation device according to one aspect of the present invention, when the normalized difference vegetation index is equal to or greater than the threshold, the correction unit may remove the third grid from the digital surface model and then interpolate the digital surface model after removing the third grid.

[0015] According to this configuration, the elevation values of the areas corresponding to the removed third grids can be corrected simply by removing from the digital surface model any third grids whose normalized difference vegetation index is equal to or greater than a threshold and performing interpolation. This allows the shape of the earth's surface to be estimated with higher accuracy and simpler processing than conventional methods.

[0016] In the estimation device according to one aspect of the present invention, when the normalized difference vegetation index is less than the threshold value, the correction unit may correct the elevation value by subtracting from the elevation value a correction value set according to the difference between the threshold value and the normalized difference vegetation index.

[0017] According to the above configuration, for the elevation values of other third grids that could not be fully corrected only by removing the third grids where the normalized difference vegetation index is equal to or greater than the threshold value, it is also possible to accurately correct them according to the vegetation state in the other third grids. As a result, the shape of the target ground surface can be grasped with higher accuracy by simpler processing than before.

[0018] A program according to an aspect of the present invention is a program for causing a computer to function as the estimation device, and is a program for causing a computer to function as the acquisition unit, the calculation unit, the generation unit, and the correction unit. According to the above configuration, it is possible to realize a program that can accurately grasp the shape of the ground surface of the estimation target with simpler processing than before.

[0019] In order to solve the above problems, an estimation method according to an aspect of the present invention is executed by one or more computers, and for a target ground surface that is a ground surface to be estimated and target vegetation that is vegetation on the target ground surface, each of a plurality of unit spaces obtained by virtually dividing them into a grid shape is used as a virtual grid, and by setting an arbitrary virtual grid among the plurality of virtual grids as a target grid, it is an estimation method for estimating the shape of the target ground surface, including: an acquisition step of at least acquiring an infrared image in which the target ground surface and the target vegetation are imaged; a calculation step of calculating a correction index indicating the vegetation state in the virtual grid corresponding to the target grid by using the first spectral intensity of the first grid corresponding to the target grid among the plurality of grids obtained by dividing the infrared image acquired in the acquisition step into a grid shape; a generation step of generating a numerical surface model as a three-dimensional model of a surface layer composed of the target ground surface and the surface of the target vegetation by using light information of reflected light reflected by the target ground surface and the target vegetation; and a correction step of correcting the elevation value of the third grid corresponding to the target grid among the plurality of grids obtained by dividing the numerical surface model generated in the generation step into a grid shape according to the comparison result between the correction index and a threshold value. According to the above configuration, it is possible to realize an estimation method that can accurately grasp the shape of the ground surface of the estimation target with simpler processing than before. [Effect of the Invention]

[0020] According to one aspect of the present invention, the shape of the ground surface can be accurately grasped with simpler processing than in the past. [Brief Description of the Drawings]

[0021]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

[0022] Hereinafter, an embodiment of the present invention will be described in detail with reference to FIGS. 1 to 11. In this embodiment, a farmland will be described as an example of the land to be estimated by the estimation device according to one aspect of the present invention. Note that the land to be estimated is not limited to farmland, and various lands such as forests, pastures, urban areas, and industrial parks can be the estimation targets.

[0023] 〔Configuration of Information Processing Apparatus〕 With reference to FIG. 1, the configuration of an information processing apparatus 1 according to an embodiment of the present invention will be described. The information processing apparatus 1 is an apparatus capable of executing processing of various kinds of information. The information processing apparatus 1 may be, for example, a tablet terminal, or may be a smartphone. Alternatively, the information processing apparatus 1 may be a desktop personal computer. As shown in FIG. 1, the information processing apparatus 1 includes an input unit 11, a display unit 12, a storage unit 13, a communication unit 14, and a control unit 15.

[0024] The input unit 11 is an interface that receives various operations. For example, when the information processing apparatus 1 is a desktop personal computer, a keyboard and a mouse serve as the input unit 11. The display unit 12 is an output unit that displays various kinds of information. For example, when the information processing apparatus 1 is a desktop personal computer, a monitor serves as the display unit 12. Also, for example, when the information processing apparatus 1 is a smartphone or a tablet terminal, the information processing apparatus 1 may include a touch panel in which the input unit 11 and the display unit 12 are integrated.

[0025] The storage unit 13 stores various kinds of information used by the information processing apparatus 1. Examples of the storage unit 13 and a storage unit 22 described later include a RAM (Random Access Memory), a flash memory, and a hard disk. The communication unit 14 is a unit for the information processing apparatus 1 to transmit and receive various kinds of information to and from a flying object 2 described later or another information processing apparatus. The control unit 15 is, for example, a CPU (Central Processing Unit) or a GPU (Graphics Processing Unit), and comprehensively controls each unit of the information processing apparatus 1. Also, the control unit 15 executes processing for realizing various functions provided in the information processing apparatus 1. As shown in FIG. 1, the control unit 15 includes an estimation device 100.

[0026] The estimation device 100 is a device that estimates the shape of the target ground surface GS. The target ground surface GS is the ground surface to be estimated by the estimation device 100. Also, it is assumed that the target ground surface GS is covered with the target vegetation VG. The target vegetation VG is the vegetation on the target ground surface GS (see FIG. 3 and the like for the above).

[0027] Here, the "ground surface" in this specification does not refer only to a surface where there is no vegetation or other things and the soil or sand is exposed throughout, or only a paved surface. Even if part or all of it is covered by a ground object, as long as the height from the reference surface (average sea level of Tokyo Bay) of the ground object is considered to be the same as the elevation at the ground surface directly below the ground object and is low enough, the surface of the ground object is also considered to be part or all of the "ground surface" in this specification. In this embodiment, the ground surface of substantially the entire area in a specific farm field is set as the target ground surface GS. Also, the land consisting of the target ground surface GS and the target vegetation VG is set as the target land LT (see the polygon surrounded by the red frame in FIG. 3 and the like).

[0028] The estimation device 100 estimates the shape of the target ground surface GS by designating an arbitrary virtual grid GK among a plurality of virtual grids GK as the target grid GT (see FIG. 5). The virtual grid GK refers to each of a plurality of unit spaces obtained by virtually dividing the target ground surface GS and the target vegetation VG into a grid pattern (see FIG. 5).

[0029] The estimation device 100 may be provided, for example, outside the control unit 15 in the information processing device 1, or may be provided on the aircraft 2. Alternatively, the estimation device 100 may be provided in another estimation device. As shown in FIG. 1, the estimation device 100 includes an acquisition unit 101, a calculation unit 102, a generation unit 103, and a correction unit 104.

[0030] The acquisition unit 101 acquires the near-infrared image Ii (see FIG. 3) of the target land LT and the optical information of the reflected light reflected by the target land. In this embodiment, as the optical information, the visible light image Iv (see FIG. 4) of the target land LT will be described as an example.

[0031] The optical information may be the distance between the laser scanner and the surface layer obtained from the time difference from when the laser scanner emits laser light to the target land LT until the reflected light of the laser light returns. The surface layer is the surface of the target land LT, in other words, the surface composed of the target ground surface GS and the surface of the target vegetation VG. Alternatively, the optical information may be the optical information of the reflected light reflected by the target land LT by infrared rays.

[0032] The calculation unit 102 calculates a correction index using the first spectral intensity and the second spectral intensity. The first spectral intensity is the spectral intensity of the first grid G1 (see FIG. 4) corresponding to the target grid GT among a plurality of grids obtained by dividing the near-infrared image Ii in a grid pattern. The second spectral intensity is the spectral intensity in a predetermined wavelength region in the second grid G2. There is no particular limitation on the predetermined wavelength region, but red in the visible region is preferable. The second grid G2 is a grid corresponding to the target grid GT included in the optical information. In the present embodiment, the second grid G2 becomes the grid corresponding to the target grid GT among the plurality of grids obtained by dividing the visible light image Iv in a grid pattern (see FIG. 4).

[0033] The correction index is an index indicating the state of vegetation in the virtual grid GK corresponding to the target grid GT. Here, the "state of vegetation" is a concept referring to a part or all of the vegetation rate, the activity of vegetation, and the like. Examples of the correction index include various vegetation indices such as the normalized difference vegetation index, the soil-adjusted vegetation index, and the vegetation enhancement index. In the present embodiment, the normalized difference vegetation index is used as the correction index. That is, in the present embodiment, the calculation unit 102 calculates the normalized difference vegetation index as the correction index by using the spectral intensity whose predetermined wavelength region is red in the visible region as the second spectral intensity. The normalized difference vegetation index is calculated by a known formula represented by the reflectance in the near-infrared region and the reflectance of red in the visible region.

[0034] The generation unit 103 generates a target DSM3 (see FIG. 6) using optical information (visible light image Iv in this embodiment). The target DSM3 is a numerical surface model generated as a three-dimensional model of the surface layer. The correction unit 104 corrects the elevation value of the third grid G3 (see FIG. 6) corresponding to the target grid GK among a plurality of grids obtained by dividing the target DSM3 into a grid pattern according to the comparison result between the correction index and the threshold value. In this embodiment, when the normalized difference vegetation index is equal to or greater than the threshold value, the correction unit 104 removes the third grid G3 from the target DSM3 and then interpolates the target DSM3 after removing the third grid G3. Details of the processing related to this interpolation will be described later.

[0035] 〔Configuration of the aircraft〕 With reference to FIG. 1, the configuration of the aircraft 2 according to an embodiment of the present invention will be described. The aircraft 2 is an aircraft capable of autonomous flight without a crew, such as a drone or a UAV (Unmanned Aerial Vehicle), and captures a visible light image Iv and a near-infrared image Ii. The aircraft 2 may be a manned aircraft such as an airplane or a helicopter. As shown in FIG. 1, the aircraft 2 includes an imaging unit 21, a storage unit 22, a communication unit 23, and a control unit 24.

[0036] The imaging unit 21 has a visible light camera and a near-infrared camera, and captures a visible light image Iv and a near-infrared image Ii simultaneously or synchronously. The storage unit 22 may record various images captured by the imaging unit 21. The communication unit 23 is a unit for the aircraft 2 to transmit and receive various information to and from the information processing device 1 or other information processing devices. The aircraft 2 transmits the visible light image Iv and the near-infrared image Ii captured by the imaging unit 21 to the information processing device 1 via the communication unit 23. The control unit 24 is, for example, a CPU or a GPU similar to the control unit 15, and comprehensively controls each part of the aircraft 2. In addition, the control unit 24 executes processing for realizing various functions provided in the aircraft 2.

[0037] Note that the imaging subject of the visible light image Iv and the near-infrared image Ii is not limited to the aircraft 2. For example, when any moving body that can move on the target ground surface GS and on the surface of the target vegetation VG is provided with the imaging unit 21 and the communication unit 23, the moving body may image at least one of the visible light image Iv and the near-infrared image Ii. In this case, the above-described moving body may transmit the captured image to the information processing apparatus 1 via the communication unit 23. Examples of the above-described moving body include a buggy, a tractor, a lawn mower, etc. to which a camera equipped with the imaging unit 21 is attached.

[0038] 〔Main processing of the estimation device〕 With reference to FIGS. 2 to 9, the main processing of the estimation device 100 will be described. As a premise, it is assumed that the aircraft 2 has transmitted each of the near-infrared image Ii shown in FIG. 3 and the visible light image Iv shown in FIG. 4 to the information processing apparatus 1. Also, the flowchart shown in FIG. 2 is merely an example, and for example, the order of each process from S11 to S14 described later may be appropriately changed.

[0039] First, in S11 (acquisition step) of the flowchart shown in FIG. 2, the acquisition unit 101 acquires the visible light image Iv and the near-infrared image Ii via the communication unit 14. Then, the acquisition unit 101 transmits the acquired visible light image Iv to each of the calculation unit 102 and the generation unit 103. Also, the acquisition unit 101 transmits the acquired near-infrared image Ii to the calculation unit 102.

[0040] Next, in S12, the estimation device 100 specifies the target grid GT. Specifically, as shown in FIG. 5, the estimation device 100 virtually divides the target land LT into a grid pattern, and each of the obtained plurality of unit spaces is set as a virtual grid GT. Then, the estimation device 100 arbitrarily extracts a virtual grid GK that is a calculation target of the calculation unit 102 and also a correction target of the correction unit 104 from among the plurality of virtual grids GK that constitute the target land LT, and sets it as the target grid GT.

[0041] Note that the division interval (grid size) of the target land LT shown in FIG. 5 is merely an example. The estimation device 100 can arbitrarily change the grid size according to the size of the target land LT, the performance (resolution, decomposition ability) of the visible light camera and the near-infrared camera that the imaging unit 21 has, the distance between the target land LT and these cameras, and the like. This also applies to the first grid G1 to the third grid G3. Further, since the photograph in FIG. 5 is an image of the target land LT taken from above, in FIG. 5, the virtual grid GK and the attention grid GT are each represented two-dimensionally (latitude and longitude). However, in reality, both the virtual grid GK and the attention grid GT are three-dimensional concepts including altitude.

[0042] Next, in S13 (generation step), the generation unit 103 generates the target DSM3 using the visible light image Iv and transmits it to the correction unit 104. In the present embodiment, the generation unit 103 generates the target DSM3 based on the three-dimensional point cloud data generated by the SfM-MVS (Structure From Motion - Multi-View stereo) process. The SfM-MVS process is a known process for creating three-dimensional point cloud data of the surface shape of a target object, and the MVS process is executed after the SfM process is executed.

[0043] Hereinafter, the details of the SfM-MVS process by the generation unit 103 will be described. First, the generation unit 103 calculates the position of the imaging unit 21 at the time of imaging from the feature points of the target land LT, which is the subject of the visible light image Iv, by executing the SfM process, and generates low-density three-dimensional point cloud data. Then, the generation unit 103 densifies the low-density three-dimensional point cloud data based on the position of the imaging unit 21 calculated in the SfM process by executing the MVS process, and generates three-dimensional point cloud data of the surface layer (hereinafter, "basic three-dimensional point cloud data") that serves as the basis for the target DSM3.

[0044] Then, the generation unit 103 divides the basic three-dimensional point cloud data into a grid pattern with a grid size preset by the user. The set value of the grid size may be stored in advance in the storage unit 13, or may be set by an input operation received by the input unit 11. Then, for all the grids obtained by dividing the basic three-dimensional point cloud data, the elevation values of each point group in the grid are statistically processed (or averaged), and the obtained average value is used as the elevation value in the grid. By aggregating the elevation values of each of the plurality of points constituting the basic three-dimensional point cloud data into the grid through these series of processes, the basic three-dimensional point cloud data is converted into the target DSM3.

[0045] Note that the generation of the target DSM3 by the SfM-MVS process is merely an example, and the generation unit 103 may generate the target DSM3 by a known process other than the SfM-MVS process. Examples of known processes other than the SfM-MVS process include a manual generation process using a digital stereogram machine and a generation process using satellite images.

[0046] Next, in S14 (calculation step), the calculation unit 102 calculates the normalized difference vegetation index using the near-infrared image Ii and the visible light image Iv, and transmits it to the correction unit 104. Specifically, as shown in FIG. 3, the calculation unit 102 designates, as the first grid G1, the grid corresponding to the target grid GT specified in the process of S12 from among the plurality of grids obtained by dividing the near-infrared image Ii into a grid pattern. Then, the calculation unit 102 acquires the reflection spectrum as the first spectral intensity from the first grid G1, and calculates the reflectance of the first grid G1 based on this reflection spectrum.

[0047] Further, as shown in FIG. 4, the calculation unit 102 sets, as the second grid G2, the grid corresponding to the target grid GT specified in the process of S12 from among a plurality of grids obtained by dividing the visible light image Iv in a grid pattern. Then, the calculation unit 102 acquires, as the second spectral intensity, a reflection spectrum in which a predetermined wavelength region is red in the visible region from the second grid G2, and calculates the reflectance of the second grid G2 based on this reflection spectrum. Then, the calculation unit 102 calculates the normalized difference vegetation index of the virtual grid GK corresponding to the target grid GT specified in the process of S12 by using the reflectance of the first grid G1 and the reflectance of the second grid G2. Thereafter, the calculation unit 102 calculates the normalized difference vegetation index of all the virtual grids GK by repeating this series of calculation processes.

[0048] Next, in S15 (correction step), the correction unit 104 compares the normalized difference vegetation index with a threshold value. As a premise, the correction unit 104 sets the threshold value by the method shown in FIG. 7. First, in S21 of the flowchart shown in FIG. 7, the correction unit 104 specifies an arbitrary third grid G3 (hereinafter, "arbitrary grid") from the target DSM3. Then, the correction unit 104 compares the normalized difference vegetation index of the virtual grid GK corresponding to the arbitrary grid with a first threshold value. The first threshold value is the upper limit value of the normalized difference vegetation index at which the surface layer of the virtual grid GK corresponding to the arbitrary grid can be regarded as the target ground surface GS, and can be arbitrarily set by the user.

[0049] If the normalized difference vegetation index of the virtual grid GK corresponding to the arbitrary grid is greater than the first threshold value (NO in S21), the correction unit 104 specifies another arbitrary grid and executes the process of S21 again. On the other hand, if the normalized difference vegetation index of the virtual grid GK corresponding to the arbitrary grid is equal to or less than the first threshold value (YES in S21), the process proceeds to S22.

[0050] In S22, the correction unit 104 calculates the difference between the normalized difference vegetation index of the virtual grid GK corresponding to an arbitrary grid and the normalized difference vegetation index of the virtual grid GK corresponding to a third grid G3 (hereinafter, "adjacent grid") adjacent to the arbitrary grid. Hereinafter, the difference between the normalized difference vegetation index of the virtual grid GK corresponding to an arbitrary grid and the normalized difference vegetation index of the virtual grid GK corresponding to the adjacent grid is referred to as the "difference in normalized difference vegetation index".

[0051] Then, the correction unit 104 compares the calculated difference in normalized difference vegetation index with a second threshold value. The second threshold value is the lower limit value of the difference in normalized difference vegetation index at which the surface layer of the virtual grid GK corresponding to the adjacent grid can be regarded as the surface of the target vegetation VG, and can be arbitrarily set by the user. If the difference in normalized difference vegetation index is less than the second threshold value (NO in S22), the correction unit 104 specifies another arbitrary grid and executes the process of S21 again. On the other hand, if the difference in normalized difference vegetation index is greater than or equal to the second threshold value (YES in S22), the process proceeds to S23.

[0052] Next, in S23, the correction unit 104 calculates the difference between the elevation value of the arbitrary grid and the elevation value of the adjacent grid (hereinafter, "difference in elevation value"). Next, in S24, the correction unit 104 determines whether the difference in elevation value has been calculated for all arbitrary grids and adjacent grids that satisfy the conditions. Here, "satisfying the conditions" means a case where the normalized difference vegetation index of the virtual grid GK corresponding to the arbitrary grid is less than or equal to the first threshold value and the difference in normalized difference vegetation index is greater than or equal to the second threshold value. If there are any arbitrary grids and adjacent grids for which the difference in elevation value has not been calculated (NO in S24), the correction unit 104 specifies another arbitrary grid and executes the process of S21 again. On the other hand, if the difference in elevation value has been calculated for all of them (YES in S24), the process proceeds to S25.

[0053] Next, in S25, the correction unit 104 searches for the elevation difference that is less than or equal to the allowable difference and is closest to the allowable difference among all the elevation differences. The allowable difference is the upper limit value of the difference between the elevation values of adjacent grids that require correction and the elevation value of any grid corresponding to the virtual grid GK whose normalized difference vegetation index is less than or equal to the first threshold. Next, in S26, the correction unit 104 extracts the elevation difference that is less than or equal to the allowable difference and is closest to the allowable difference, and sets the normalized difference vegetation index corresponding to the extracted elevation difference as the threshold. Here, the "normalized difference vegetation index corresponding to the elevation difference" refers to the normalized difference vegetation index of the virtual grid GK corresponding to the target adjacent grid. The "target adjacent grid" is an adjacent grid for which the elevation difference that is less than or equal to the allowable difference and is closest to the allowable difference is obtained. With the completion of the process in S26, all the threshold setting processes executed by the correction unit 104 are completed.

[0054] Hereinafter, with reference to FIG. 8, a specific example of the threshold setting method shown in FIG. 7 will be described. In the example of FIG. 8, it is assumed that the first threshold is set to 0.1, the second threshold is set to 0.3, and the allowable difference is set to 0.2 m. In the example of FIG. 8, the arbitrary grids corresponding to the virtual grid GK whose normalized difference vegetation index is less than or equal to the first threshold are two grids, namely, the third grid G3 with a normalized difference vegetation index of 0.0 and the third grid G3 with a normalized difference vegetation index of 0.1.

[0055] Regarding the third grid G3 with a normalized difference vegetation index of 0.0 of the virtual grid GK and the third grid G3 with a normalized difference vegetation index of 0.6 of the adjacent virtual grid GK, the difference in the normalized difference vegetation index is 0.6, which is greater than or equal to the second threshold. However, since the elevation difference between these two third grids G3 is greater than the allowable difference, 0.6 is excluded from the threshold candidates. Regarding the third grid G3 with a normalized difference vegetation index of 0.1 of the virtual grid GK and the third grid G3 with a normalized difference vegetation index of 0.8 of the adjacent virtual grid GK, the difference in the normalized difference vegetation index is 0.7, which is greater than or equal to the second threshold. However, since the elevation difference between these two third grids G3 is also greater than the allowable difference, 0.8 is also excluded from the threshold candidates.

[0056] For the third grid G3 with a normalized difference vegetation index of the virtual grid GK being 0.1 and the third grid G3 with a normalized difference vegetation index of an adjacent virtual grid GK being 0.4, the difference in the normalized difference vegetation index is 0.3, which is equal to or greater than the second threshold. Here, since the difference in elevation values between these two third grids G3 is smaller than the allowable difference, 0.4 is a candidate for the threshold. And in the example of FIG. 8, since the only adjacent grid with a difference in elevation value equal to or less than the allowable difference is the adjacent grid with a normalized difference vegetation index of 0.4, the target adjacent grid automatically becomes the adjacent grid with a normalized difference vegetation index of 0.4. Therefore, in the example of FIG. 8, the threshold is 0.4.

[0057] Note that the method for setting the threshold is not limited to the foregoing example. For example, the threshold may be stored in advance in the memory of the storage unit 13 or the correction unit 104, or the input unit 11 may receive a user operation related to threshold setting to set the threshold.

[0058] When the normalized difference vegetation index is equal to or greater than the threshold (YES in S15), the process proceeds to S16. In S16 (correction step), the correction unit 104 removes the third grid G3. Specifically, as shown in FIG. 6, the correction unit 104 sets the grid corresponding to the target grid GT specified in the process of S12 as the third grid G3 among the plurality of grids obtained by dividing the target DSM3 in a grid pattern. Then, the correction unit 104 removes the third grid G3 from the target DSM3. When the process of S16 ends, the calculation unit 102 executes the process of S14 again.

[0059] On the other hand, when the normalized difference vegetation index is less than the threshold (NO in S15), the process proceeds to S17. In S17, the correction unit 104 determines whether the comparison with the threshold has been completed for the normalized difference vegetation index corresponding to each of the plurality of virtual grids GK that make up the target land LT.

[0060] When it is determined that not all comparisons with the threshold value have been completed (NO in S17), the calculation unit 102 executes the process of S14 again. On the other hand, when it is determined that all comparisons with the threshold value have been completed (YES in S17), the process proceeds to S18. Here, when all comparisons between the normalized differential vegetation index and the threshold value are completed and all the third grids G3 where the normalized differential vegetation index is greater than or equal to the threshold value are removed, the target DSM3 becomes a state like the target DSM3' shown in FIG. 9. Specifically, in FIG. 9, the green area is the area where the normalized differential vegetation index is less than the threshold value, and the purple area in FIG. 9 is the area where the normalized differential vegetation index is greater than or equal to the threshold value (that is, the area where the third grid G8 is removed).

[0061] In S18 (correction step), the correction unit 104 interpolates the target DSM3'. Specifically, the correction unit 104 extracts the elevation points of all the third grids G3 remaining in the target DSM3'. Then, the correction unit 104 interpolates between two adjacent elevation points by a known interpolation method such as the kriging method. Known interpolation methods other than the kriging method include the spline method, the trend method, the natural neighbor method, etc.

[0062] The target DSM3' after the interpolation is completed becomes a state like the target DSM3'' shown in FIG. 10. When the target DSM3'' shown in FIG. 10 is compared with the comparison DSM4 shown in FIG. 11, the two are very similar. The comparison DSM4 is a DSM in a state where the target land LT is composed of only the approximate target ground surface GS. In other words, it is approximately equivalent to the DEM obtained by performing filtering processing on the target DSM3. With the end of the process of S18, all the main processes executed by the estimation device 100 are completed.

[0063] 〔Modification Example〕 A modification of the present invention will be described below. First, in the present embodiment, various processes of the estimation device 100 are executed by one computer, but these various processes may be executed by a plurality of computers. In other words, all components (acquisition unit 101, calculation unit 102, generation unit 103, correction unit 104) of the estimation device 100 do not have to be provided in the one estimation device 100, and some components may be provided in other information processing devices (computers). In this case, an estimation system can be constructed by the information processing device 1, other information processing devices, and the aircraft 2. Further, a part or all of the near-infrared image Ii, visible light image Iv, target DSM3, and target DSM3'' may be stored in a database and stored and managed by a server. In these cases, an estimation system can be constructed by the information processing device 1, the server, and the aircraft 2.

[0064] Next, the acquisition unit 101 does not have to acquire optical information. In this case, the calculation unit 102 calculates another correction index instead of the normalized difference vegetation index using only the first spectral intensity acquired from the first grid G1 of the near-infrared image Ii. Examples of other correction indices instead of the normalized difference vegetation index include the normalized difference NIR / MIR modified normalized difference vegetation index, the normalized difference moisture index, and the like. Also in this case, for example, the calculation unit 102 directly acquires optical information. Further, the acquisition unit 101 may acquire a short-wave infrared image, a mid-infrared image, or a far-infrared image instead of the near-infrared image Ii. When the acquisition unit 101 acquires a far-infrared image, the calculation unit 102 acquires a radiation spectrum as the first spectral intensity from the first grid of the far-infrared image.

[0065] In other words, the acquisition unit 101 only needs to acquire at least an infrared image. Further, the calculation unit 102 only needs to calculate a correction index using the first spectral intensity of the first grid corresponding to the target grid GT among a plurality of grids obtained by dividing the infrared image into a grid pattern.

[0066] Next, when the normalized differential vegetation index is less than the threshold value, the correction unit 104 may correct the elevation value of the third grid G3 by subtracting a correction value set according to the difference between the threshold value and the normalized differential vegetation index (hereinafter abbreviated as "difference") from the elevation value of the third grid G3. For example, if the difference is 0.1, the correction value may be set to 0.9 m; if the difference is 0.2, the correction value may be set to 0.6 m; if the difference is 0.3, the correction value may be set to 0.3 m; and if the difference is 0.4, the correction value may be set to 0.0 m (no correction).

[0067] According to such a configuration, not only the removal of the third grid G3 where the normalized differential vegetation index is equal to or greater than the threshold value, but also the elevation values of other third grids G3 that could not be fully corrected can be accurately corrected according to the vegetation state in the other third grids G3. As a result, the shape of the target ground surface GS can be grasped with higher accuracy by a simpler process than before.

[0068] 〔Example of Realization by Software〕 The functions of the estimation device 100 (hereinafter abbreviated as "device") can be realized by a program for causing a computer to function as the device, and by a program for causing a computer to function as each control block of the device (especially each part included in the control unit 15).

[0069] In this case, the device includes a computer having at least one control device (for example, a processor) and at least one storage device (for example, a memory) as hardware for executing the program. By executing the program with this control device and storage device, each function described in each of the above embodiments is realized.

[0070] The program may be recorded on one or more computer-readable recording media, not temporarily. This recording medium may or may not be provided in the device. In the latter case, the program may be supplied to the device via any wired or wireless transmission medium.

[0071] In addition, part or all of the functions of each of the control blocks can also be realized by a logic circuit. For example, an integrated circuit in which a logic circuit functioning as each of the control blocks is formed is also included in the scope of the present invention. In addition to this, for example, it is also possible to realize the functions of each of the control blocks by a quantum computer.

[0072] 〔Supplementary Note〕 The present invention is not limited to the above-described embodiments and modified examples, and various modifications are possible within the scope shown in the claims. Embodiments obtained by appropriately combining the technical means disclosed in different embodiments are also included in the technical scope of the present invention.

Explanation of Reference Numerals

[0073] 3 Target DSM (Digital Surface Model) 3´ Target DSM (Digital Surface Model after removing the third grid) 100 Estimation device 101 Acquisition unit 102 Calculation unit 103 Generation unit 104 Correction unit GS Target ground surface VG Target vegetation G1 First grid G2 Second grid G3 Third grid GK Virtual grid GT Target grid Ii Near-infrared image Iv Visible light image (optical information)

Claims

1. An estimation device for estimating the shape of a target ground surface, which is the ground surface to be estimated, and target vegetation, which is the vegetation on the target ground surface, by using, as virtual grids, each of a plurality of unit spaces obtained by virtually dividing the target ground surface and the target vegetation into a grid pattern, and designating an arbitrary virtual grid among the plurality of virtual grids as a target grid, comprising: an acquisition unit configured to acquire at least an infrared image of the target ground surface and the target vegetation; a calculation unit configured to calculate a correction index indicating a vegetation state in the virtual grid corresponding to the target grid by using a first spectral intensity of a first grid corresponding to the target grid among a plurality of grids obtained by dividing the infrared image into a grid pattern; a generation unit configured to generate a numerical surface model as a three-dimensional model of a surface layer formed by the target ground surface and a surface of the target vegetation by using light information of reflected light reflected by the target ground surface and the target vegetation; a correction unit configured to correct an elevation value of a third grid corresponding to the target grid according to a comparison result between the correction index and a threshold value among a plurality of grids obtained by dividing the numerical surface model into a grid pattern.

2. The acquisition unit acquires a near-infrared image as the infrared image and further acquires the light information, The calculation unit, uses the first spectral intensity, and a second spectral intensity in a predetermined wavelength range in a second grid corresponding to the target grid included in the light information, to calculate the correction index. The estimation device according to claim 1.

3. The calculation unit calculates a normalized difference vegetation index as the correction index by using, as the second spectral intensity, a spectral intensity in which the predetermined wavelength range is red in the visible region, The correction unit corrects the elevation value according to a comparison result between the normalized difference vegetation index and the threshold value. The estimation device according to claim 2.

4. When the normalized difference vegetation index is equal to or greater than the threshold value, the correction unit removes the third grid from the numerical surface model and then interpolates the numerical surface model after removing the third grid. The estimation device according to claim 3.

5. When the normalized difference vegetation index is less than the threshold value, the correction unit corrects the elevation value by subtracting a correction value set according to a difference between the threshold value and the normalized difference vegetation index from the elevation value. The estimation device according to claim 3.

6. A program for causing a computer to function as the estimation device according to any one of Claims 1 to 5, the program for causing a computer to function as the acquisition unit, the calculation unit, the generation unit, and the correction unit.

7. It is executed by one or more computers, and Regarding a target ground surface that is a ground surface to be estimated and target vegetation that is vegetation on the target ground surface, each of a plurality of unit spaces obtained by virtually dividing the target ground surface and the target vegetation in a grid pattern is used as a virtual grid, and an arbitrary virtual grid among the plurality of virtual grids is used as a target grid. An estimation method for estimating the shape of the target ground surface, An acquisition step of acquiring at least an infrared image in which the target ground surface and the target vegetation are imaged; A calculation step of calculating a correction index indicating the state of vegetation in the virtual grid corresponding to the target grid by using the first spectral intensity of the first grid corresponding to the target grid among a plurality of grids obtained by dividing the infrared image acquired in the acquisition step in a grid pattern; A generation step of generating a numerical surface model as a three-dimensional model of a surface layer composed of the target ground surface and the surface of the target vegetation by using light information of reflected light reflected by the target ground surface and the target vegetation; A correction step of correcting the elevation value of the third grid corresponding to the target grid among a plurality of grids obtained by dividing the numerical surface model generated in the generation step in a grid pattern according to the comparison result between the correction index and a threshold value.

Images