Forest resource information calculation device, forest resource information calculation method, and program

Abstract

To properly analyze broad-leaved trees included in a survey target forest area at a single-tree level.SOLUTION: A forest resource information calculation device creates three-dimensional point cloud data in which influence of inclination of a ground surface is excluded from laser measurement data obtained by irradiating a survey target forest area with laser light from multiple directions, creates a plurality of pieces of slice data by dividing the three-dimensional point cloud data in a height direction, and performs single-tree analysis of broad-leaved trees included in the survey target forest area. In the single-tree analysis, a tree trunk point cloud cluster is generated for each piece of slice data by clustering the three-dimensional point cloud data included in each piece of slice data, stacking of the tree trunk point cloud cluster included in one piece of slice data and the tree trunk point cloud cluster included in adjacent piece of slice data is performed over the entire three-dimensional point cloud data in the height direction, and determination is made as to whether the tree trunk point cloud cluster including a branching portion corresponds to two different broad-leaved trees or to one broad-leaved tree having a branch branching from the trunk.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to a forest resource information calculation device, a forest resource information calculation method, and a program. [Background technology]

[0002] Forests cover 70% of the country's land area, 50% of which is broadleaf forest. Hardwood lumber is used for furniture, flooring, decorative materials, etc., and is traded at prices several to ten times higher than those of softwood. However, 90% of the demand for hardwood lumber is imported. More than 90% of domestically produced hardwood is used as a cheap chip material. Although hardwood resources are abundant, they are currently not being utilized despite their high value. The market value of hardwood is determined by the tree species, tree shape (the presence or absence of trunk curvature and branching), and the amount of lumber produced. Domestic furniture and housing manufacturers are not informed of where and how much hardwood resources are in the country's forests, and so a domestic hardwood market has not been formed.

[0003] In recent years, techniques for calculating forest resource information have been proposed (see, for example, Patent Document 1). In the technique described in Patent Document 1, image data corresponding to a forest area to be surveyed is acquired by aerial photography, and three-dimensional point cloud data is created based on the aerial photography image data. Furthermore, image data of the forest area to be surveyed is created based on the three-dimensional point cloud data and geographical information obtained from a geographic information system. Also, based on the forest area to be surveyed image data, canopy height image data indicating the canopy height is created. Furthermore, processing for calculating various forest resource information is performed by using the canopy height image data. Incidentally, in the technology described in Patent Document 1, it is assumed that the trees to be surveyed are tall planted trees (coniferous trees) such as larch, red pine, cedar, and cypress. When the trees to be surveyed are coniferous trees, it is easier to find the tree apex compared to when the trees to be surveyed are broad-leaved trees. In addition, coniferous trees have fewer trunk bends and branches compared to broad-leaved trees. Therefore, in the technology described in Patent Document 1, in order to calculate various forest resource information, image data created based on aerial image data is basically used, and low-density three-dimensional point cloud data (e.g., data including tree crown height, ground surface elevation, etc.) created using SFM (Structure from Motion) or laser measurement data is used as a supplement. Therefore, the technology described in Patent Document 1 cannot properly analyze broadleaf trees included in the surveyed forest area at the individual tree level, and cannot properly calculate forest resource information related to broadleaf trees (e.g., information indicating the tree shape, growth status, etc. of broadleaf trees). [Prior art documents] [Patent documents]

[0004] [Patent Document 1] JP 2019-185449 A Summary of the Invention [Problem to be solved by the invention]

[0005] In consideration of the above-mentioned problems, the present invention aims to provide a forest resource information calculation device, a forest resource information calculation method, and a program that can appropriately analyze broad-leaved trees contained in a surveyed forest area at the individual tree level. [Means for solving the problem]

[0006] One aspect of the present invention includes a laser measurement data acquisition unit that acquires laser measurement data obtained by irradiating a survey target forest area with laser light from multiple directions; a three-dimensional point cloud data creation unit that creates three-dimensional point cloud data based on the laser measurement data acquired by the laser measurement data acquisition unit; a three-dimensional point cloud data correction unit that creates corrected three-dimensional point cloud data, which is three-dimensional point cloud data obtained by excluding the influence of the inclination of the ground surface from the three-dimensional point cloud data created by the three-dimensional point cloud data creation unit; a slice data creation unit that creates a plurality of slice data by dividing the corrected three-dimensional point cloud data created by the three-dimensional point cloud data correction unit in the height direction; and an individual tree analysis unit that analyzes each of a plurality of broad-leaved trees included in the survey target forest area, wherein the individual tree analysis unit performs clustering of the 3D point cloud data included in each of the plurality of slice data created by the slice data creation unit. a first clustering unit that generates, for each slice data, a trunk point cloud cluster corresponding to each of the trunks of a plurality of broad-leaved trees included in the forest area to be surveyed; a second clustering unit that generates a trunk point cloud cluster corresponding to each of the trunks of a plurality of broad-leaved trees included in the forest area to be surveyed by stacking a plurality of trunk point cloud clusters included in one slice data with a plurality of trunk point cloud clusters included in adjacent slice data, which is slice data adjacent to the one slice data, over the entire height direction of the corrected 3D point cloud data; and a separation processing unit that, when a trunk point cloud cluster including a branching portion is generated by the second clustering unit, determines whether the trunk point cloud cluster including a branching portion corresponds to two different broad-leaved trees or whether the trunk point cloud cluster including a branching portion corresponds to a single broad-leaved tree having a trunk and a branch branching off from the trunk.

[0007] In one embodiment of the forest resource information calculation device of the present invention, the separation processing unit may determine, based on the height dimension of the branching portion, whether the trunk point cloud cluster including the branching portion corresponds to two different broad-leaved trees, or whether the trunk point cloud cluster including the branching portion corresponds to a single broad-leaved tree.

[0008] In one embodiment of the forest resource information calculation device of the present invention, the separation processing unit may determine that the trunk point cloud cluster including the branching portion corresponds to two different broad-leaved trees if the height dimension of the branching portion is more than two-thirds of the height dimension of the corrected 3D point cloud data created by the 3D point cloud data correction unit, and may determine that the trunk point cloud cluster including the branching portion corresponds to a single broad-leaved tree if the height dimension of the branching portion is less than two-thirds of the height dimension of the corrected 3D point cloud data created by the 3D point cloud data correction unit.

[0009] In one embodiment of the forest resource information calculation device of the present invention, the single tree analysis unit may be provided with a trunk center point candidate calculation unit that calculates candidates for the center point of the trunk of a broad-leaved tree based on the point cloud contained in the trunk point cloud cluster generated by the second clustering unit.

[0010] In one embodiment of the forest resource information calculation device of the present invention, the single tree analysis unit may be provided with a trunk center point determination unit that determines whether the candidate trunk center point calculated by the trunk center point candidate calculation unit is a trunk center point based on the number of candidate trunk center points calculated by the trunk center point candidate calculation unit.

[0011] In one embodiment of the forest resource information calculation device of the present invention, the single tree analysis unit may be provided with an individual tree judgment unit that determines whether the trunk point cluster to be judged corresponds to each of the trunks of a plurality of broad-leaved trees included in the surveyed forest area, or is noise, based on the point judged to be a trunk center point by the trunk center point judgment unit and the result of the judgment by the separation processing unit.

[0012] A forest resource information calculation device of one embodiment of the present invention may be provided with a single tree point cloud extraction unit that reproduces parts other than the trunks of multiple broad-leaved trees included in the surveyed forest area by using 3D point cloud data created by the 3D point cloud data creation unit, other than 3D point cloud data that has been determined by the single tree judgment unit to be a trunk point cloud cluster corresponding to a tree trunk.

[0013] A forest resource information calculation device of one embodiment of the present invention may be provided with an individual tree height analysis unit that calculates the horizontal position of each of a number of broad-leaved trees included in the surveyed forest area by using 3D point cloud data that has been determined by the individual tree judgment unit to be a trunk point cloud cluster corresponding to a tree trunk, and 3D point cloud data that has been used by the individual tree point cloud extraction unit to reproduce parts other than the trunk.

[0014] One aspect of the present invention includes a laser measurement data acquisition step of acquiring laser measurement data obtained by irradiating a forest area to be surveyed with laser light from multiple directions, a three-dimensional point cloud data creation step of creating three-dimensional point cloud data based on the laser measurement data acquired in the laser measurement data acquisition step, a three-dimensional point cloud data correction step of creating corrected three-dimensional point cloud data, which is three-dimensional point cloud data obtained by excluding the influence of the inclination of the earth's surface from the three-dimensional point cloud data created in the three-dimensional point cloud data creation step, a slice data creation step of creating a plurality of slice data by dividing the corrected three-dimensional point cloud data created in the three-dimensional point cloud data correction step in the height direction, and an individual tree analysis step of analyzing each of a plurality of broad-leaved trees included in the forest area to be surveyed, wherein the individual tree analysis step includes a cluster analysis of the 3D point cloud data included in each of the plurality of slice data created in the slice data creation step. a first clustering step of generating trunk point cloud clusters corresponding to the trunks of each of a plurality of broad-leaved trees included in the forest area to be surveyed for each slice data by stacking the plurality of trunk point cloud clusters included in one slice data with the plurality of trunk point cloud clusters included in adjacent slice data, which is slice data adjacent to the one slice data, over the entire height direction of the corrected 3D point cloud data, and a separation processing step of determining, when a trunk point cloud cluster including a branching portion is generated in the second clustering step, whether the trunk point cloud cluster including a branching portion corresponds to two different broad-leaved trees or whether the trunk point cloud cluster including a branching portion corresponds to a single broad-leaved tree having a trunk and a branch branching off from the trunk.

[0015] One aspect of the present invention is a program for causing a computer to execute a laser measurement data acquisition step of acquiring laser measurement data obtained by irradiating a survey target forest area with laser light from multiple directions, a three-dimensional point cloud data creation step of creating three-dimensional point cloud data based on the laser measurement data acquired in the laser measurement data acquisition step, a three-dimensional point cloud data correction step of creating corrected three-dimensional point cloud data, which is three-dimensional point cloud data obtained by excluding the influence of the inclination of the earth's surface from the three-dimensional point cloud data created in the three-dimensional point cloud data creation step, a slice data creation step of creating a plurality of slice data by dividing the corrected three-dimensional point cloud data created in the three-dimensional point cloud data correction step in the height direction, and an individual tree analysis step of analyzing each of a plurality of broad-leaved trees included in the survey target forest area, wherein the individual tree analysis step includes a step of analyzing three-dimensional data included in each of the plurality of slice data created in the slice data creation step. the first clustering step of generating, for each slice data, a trunk point cloud cluster corresponding to each trunk of a plurality of broad-leaved trees included in the forest area to be surveyed by clustering the three-dimensional point cloud data; a second clustering step of generating a trunk point cloud cluster corresponding to each trunk of a plurality of broad-leaved trees included in the forest area to be surveyed by stacking a plurality of trunk point cloud clusters included in one slice data with a plurality of trunk point cloud clusters included in adjacent slice data, which is slice data adjacent to the one slice data, over the entire height direction of the corrected three-dimensional point cloud data; and a separation processing step of, when a trunk point cloud cluster including a branching portion is generated in the second clustering step, determining whether the trunk point cloud cluster including a branching portion corresponds to two different broad-leaved trees or whether the trunk point cloud cluster including a branching portion corresponds to a single broad-leaved tree having a trunk and a branch branching off from the trunk. Effect of the Invention

[0016] According to the present invention, it is possible to provide a forest resource information calculation device, a forest resource information calculation method, and a program that can appropriately analyze broad-leaved trees contained in a surveyed forest area at the individual tree level. [Brief description of the drawings]

[0017] [Figure 1] 1 is a diagram illustrating an example of a forest resource information calculation device according to a first embodiment. [Diagram 2] 3 is a diagram showing an example of three-dimensional point cloud data created by a three-dimensional point cloud data creation unit; FIG. [Diagram 3] 3 is a diagram showing an example of three-dimensional point cloud data created by a three-dimensional point cloud data creation unit; FIG. [Figure 4] 11 is a diagram for explaining an example of processing performed by a three-dimensional point cloud data correction unit. FIG. [Diagram 5] 11 is a diagram showing an example of a relationship between corrected three-dimensional point cloud data created by a three-dimensional point cloud data correcting unit and a plurality of slice data created by a slice data creating unit. FIG. [Figure 6] 11 is a diagram showing an example of a plurality of slice data created by a slice data creating unit; FIG. [Figure 7] FIG. 2 is a diagram for explaining clustering performed by a first clustering unit. [Figure 8] FIG. 13 is a diagram showing an example of a plurality of trunk point cloud clusters corresponding to the trunks of a plurality of broad-leaved trees generated by a second clustering unit. [Figure 9] 13 is a diagram for explaining an example in which a separation processing unit determines that a trunk point cloud cluster including a branch portion corresponds to two different broad-leaved trees. FIG. [Figure 10] 13 is a diagram showing an example of a trunk point cloud cluster that has been determined by the individual tree determination unit to correspond to the trunks of multiple broad-leaved trees contained in a survey target forest area. FIG. [Figure 11] 13 is a diagram showing an example of 3D point cloud data of multiple broad-leaved trees included in a surveyed forest area after parts other than the trunks have been reproduced by the single tree point cloud extraction unit. FIG. [Figure 12]12 is a diagram showing data in which information indicating the horizontal position and tree height calculated by a single tree height analysis unit is added to the three-dimensional point cloud data shown in FIG. 11. FIG. [Figure 13] 4 is a flowchart for explaining an example of processing executed in the forest resource information calculation device of the first embodiment. [Figure 14] 14 is a flowchart for explaining in detail an example of the process executed in step S50 of FIG. 13. [Figure 15] This is a diagram to explain the forest area surveyed in Omachi City, the first survey site. [Figure 16] This is a diagram to explain the forest area surveyed in Futatsuya, Omachi City, the second survey site. [Figure 17] This is a diagram to explain the forest area surveyed in Kijimadaira Village, the third survey site. [Figure 18] This figure shows the results of field surveys at the three study sites. [Figure 19] This shows the classification results for the first and second survey sites. [Figure 20] This figure shows the classification results of the third survey site. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0018] Hereinafter, embodiments of the forest resource information calculation apparatus, the forest resource information calculation method, and the program of the present invention will be described.

[0019] First Embodiment FIG. 1 is a diagram showing an example of a forest resource information calculation device 1 according to the first embodiment. 1, a forest resource information calculation device 1 calculates forest resource information (e.g., information indicating the tree shape, growth status, etc. of each of a plurality of broad-leaved trees included in a forest area to be surveyed) in a predetermined forest area to be surveyed. The forest resource information calculation device 1 includes a laser measurement data acquisition unit 11, a 3D point cloud data creation unit 12, a 3D point cloud data correction unit 13, a slice data creation unit 14, a single tree analysis unit 15, a single tree point cloud extraction unit 16, and a single tree height analysis unit 17. The laser measurement data acquisition unit 11 acquires laser measurement data obtained by irradiating a survey target forest area with laser light from multiple directions using a drone. The laser measurement data acquired by the laser measurement data acquisition unit 11 includes, for example, GNSS (Global Navigation Satellite System) observation data of electronic reference points, GNSS observation data on the drone, IMU (Inertial Measurement Unit) observation data, laser ranging data, and the like. In the example shown in Figure 1, laser measurement data obtained by using a drone to irradiate the forest area to be surveyed with laser light from multiple directions is acquired by the laser measurement data acquisition unit 11, but in other examples, laser measurement data obtained by using a drone to irradiate the forest area to be surveyed with laser light from multiple directions and laser measurement data obtained by using a ground-based laser measurement device to irradiate the forest area to be surveyed with laser light from multiple directions may be acquired by the laser measurement data acquisition unit 11. In yet another example, the laser measurement data acquisition unit 11 may acquire laser measurement data obtained by irradiating the forest area being surveyed with laser light from multiple directions using only a ground-based laser measurement device, or the laser measurement data acquisition unit 11 may acquire laser measurement data obtained by irradiating the forest area being surveyed with laser light from multiple directions using only an aircraft or helicopter, or these methods may be combined to acquire laser measurement data by the laser measurement data acquisition unit 11. In the example shown in FIG. 1, the three-dimensional point cloud data creation unit 12 creates three-dimensional point cloud data based on the laser measurement data acquired by the laser measurement data acquisition unit 11.

[0020] Figures 2 and 3 are diagrams showing an example of the 3D point cloud data created by the 3D point cloud data creation unit 12. In detail, Figure 2 shows the 3D point cloud data created by the 3D point cloud data creation unit 12 in a perspective view, Figure 3(A) shows a reduced version of the perspective view shown in Figure 2, and Figure 3(B) shows a vertical cross section of a part of the surveyed forest area shown in Figure 3(A). In other words, Figure 3(B) shows that the ground surface is inclined. As shown in FIGS. 2 and 3, in the forest resource information calculation device 1 of the first embodiment, three-dimensional point cloud data with significantly higher density is created as compared to the technique described in Patent Document 1. In addition, in the first embodiment of the forest resource information calculation device 1, laser measurement data obtained by using a drone to shine laser light onto the forest area being surveyed from multiple directions (more specifically, from multiple directions) is used, so that three-dimensional point cloud data (see Figure 3(B)) of positions that are blind spots from the angle shown in Figure 3(A) is also created.

[0021] In the example shown in Figure 1, the 3D point cloud data correction unit 13 creates corrected 3D point cloud data, which is 3D point cloud data that excludes the effects of the inclination of the earth's surface, from the 3D point cloud data created by the 3D point cloud data creation unit 12.

[0022] Fig. 4 is a diagram for explaining an example of processing performed by the three-dimensional point cloud data correction unit 13. In detail, Fig. 4(A) shows an example of three-dimensional point cloud data before processing by the three-dimensional point cloud data correction unit 13 (i.e., three-dimensional point cloud data created by the three-dimensional point cloud data creation unit 12), and Fig. 4(B) shows an example of three-dimensional point cloud data after processing by the three-dimensional point cloud data correction unit 13 (i.e., corrected three-dimensional point cloud data created by the three-dimensional point cloud data correction unit 12). When the three-dimensional point cloud data (see Figure 4(A)) created by the three-dimensional point cloud data creation unit 12 includes an inclined ground surface, the three-dimensional point cloud data correction unit 13 corrects the three-dimensional point cloud data created by the three-dimensional point cloud data creation unit 12 so that the ground surface included in the corrected three-dimensional point cloud data (see Figure 4(B)) becomes a horizontal plane (i.e., flat at the same altitude).

[0023] In the example shown in FIG. 1, the slice data generating unit 14 generates a plurality of slice data by dividing the corrected three-dimensional point cloud data generated by the three-dimensional point cloud data correcting unit 13 in the height direction.

[0024] FIG. 5 is a diagram showing an example of the relationship between the corrected three-dimensional point cloud data created by the three-dimensional point cloud data correcting unit 13 and a plurality of slice data created by the slice data creating unit 14. As shown in FIG. 5, the height dimension of the corrected 3D point cloud data created by the 3D point cloud data correcting unit 13 is about 20 m (more specifically, 20 m20 cm). Also, the slice data creating unit 14 divides the portion of the corrected 3D point cloud data from 1 m to 20 m20 cm in height at 20 cm intervals to create 96 slice data.

[0025] Fig. 6 is a diagram showing an example of a plurality of slice data created by the slice data creation unit 14. In detail, Fig. 6 shows six slice data out of the plurality of slice data created by the slice data creation unit 14. In Figure 6, "1 st For example, "3 Slice" indicates the slice data (slice data with a height of 1 m to 1.2 m) located at the bottom in the height direction (vertical direction in FIG. 5) of the corrected 3D point cloud data among the multiple slice data as shown in FIG. 5. rd For example, "5 Slice" indicates the slice data located third from the bottom in the height direction of the corrected 3D point cloud data among the multiple slice data as shown in FIG. 5 (slice data with a height of 1.40m to 1.60m). th"Slice" indicates, for example, the slice data (slice data with a height of 1.80 m to 2 m) located fifth from the bottom in the height direction of the corrected three-dimensional point cloud data among a plurality of slice data as shown in FIG.

[0026] In the example shown in Fig. 1, the single tree analysis unit 15 analyzes each of a plurality of broad-leaved trees contained in the surveyed forest area (see Fig. 2). The single tree analysis unit 15 includes a first clustering unit 15A, a second clustering unit 15B, a trunk center point candidate calculation unit 15C, a trunk center point determination unit 15D, a separation processing unit 15E, and a single tree determination unit 15F. The first clustering unit 15A performs clustering of the three-dimensional point cloud data included in each of the slice data created by the slice data creation unit 14, thereby generating a trunk point cloud cluster corresponding to each of the trunks of a plurality of broad-leaved trees included in the survey target forest area for each slice data. In detail, the first clustering unit 15A performs clustering of the three-dimensional point cloud data included in each slice data by using DBSCAN (Density-based spatial clustering of applications with noise).

[0027] Fig. 7 is a diagram for explaining the clustering performed by the first clustering unit 15A. In detail, Fig. 7(A) shows slice data with heights of 1m to 1m20cm before clustering is performed by the first clustering unit 15A (i.e., slice data with heights of 1m to 1m20cm created by the slice data creation unit 14). Fig. 7(B) shows a plurality of trunk point clusters with heights of 1m to 1m20cm after clustering is performed by the first clustering unit 15A. As shown in Fig. 7(A), the slice data before clustering by the first clustering unit 15A includes a point cloud reflecting a laser beam emitted from a drone. That is, the slice data before clustering by the first clustering unit 15A includes not only a point cloud corresponding to the trunk of a broad-leaved tree, but also a point cloud corresponding to twigs, leaves, etc. In addition, in the slice data before clustering by the first clustering unit 15A, it is not possible to identify which point cloud corresponds to one broad-leaved tree.

[0028] As shown in Fig. 7(B), the point cloud data after clustering by the first clustering unit 15A does not include point clouds corresponding to twigs, leaves, etc. In detail, the first clustering unit 15A determines that point clouds (low-density point clouds) in which the number Kn of adjacent point clouds is equal to or less than the threshold Kr among the point clouds shown in Fig. 7(A) are point clouds corresponding to twigs, leaves, etc. Furthermore, the first clustering unit 15A removes the point clouds corresponding to twigs, leaves, etc. from the point cloud data shown in Fig. 7(B). Furthermore, the clustering performed by the first clustering unit 15A identifies points corresponding to the trunks of the same (single) broad-leaved tree. Specifically, a first trunk point cloud cluster is generated by the points corresponding to the trunk of the first broad-leaved tree, and a second trunk point cloud cluster is generated by the points corresponding to the trunk of a second broad-leaved tree different from the first broad-leaved tree. The first trunk point cloud cluster and the second trunk point cloud cluster are identified as different trunk point cloud clusters.

[0029] In the example shown in Figure 1, the second clustering unit 15B stacks multiple trunk point cloud clusters contained in one slice data with multiple trunk point cloud clusters contained in adjacent slice data, which is the slice data adjacent to that slice data, throughout the entire height direction of the corrected 3D point cloud data (up and down direction in Figure 5), to generate trunk point cloud clusters corresponding to each of the trunks of multiple broad-leaved trees contained in the surveyed forest area (see Figure 1).

[0030] FIG. 8 is a diagram showing an example of a plurality of trunk point clusters corresponding to the trunks of a plurality of broad-leaved trees generated by the second clustering unit 15B. The second clustering unit 15B stacks multiple trunk point cloud clusters (stem point cloud clusters generated by the first clustering unit 15A) contained in slice data with heights of 1m to 1m20cm (the slice data located second from the bottom in Figure 5) on multiple trunk point cloud clusters (stem point cloud clusters generated by the first clustering unit 15A) contained in slice data with heights of 1m to 1m20cm (the slice data located at the bottommost part of Figure 5). Next, the second clustering unit 15B further stacks multiple trunk point cloud clusters (stem point cloud clusters generated by the first clustering unit 15A) contained in the slice data with heights of 140 cm to 160 cm (the slice data located third from the bottom in Figure 5). Next, the second clustering unit 15B further stacks multiple trunk point cloud clusters (stem point cloud clusters generated by the first clustering unit 15A) contained in the slice data with heights of 160 cm to 180 cm (the slice data located fourth from the bottom in Figure 5). The second clustering unit 15B continues stacking until multiple trunk point cloud clusters (trunk point cloud clusters generated by the first clustering unit 15A) contained in the slice data with heights of 20 m to 20 m20 cm (the slice data located at the top of Figure 5) are stacked. As a result, as shown in FIG. 8, multiple trunk point cloud clusters are generated that are stacked in the height direction (the vertical direction in FIG. 5). The trunk point cloud cluster shown in Figure 8 (the trunk point cloud cluster generated by the second clustering unit 15B) is a point cloud reflected by laser light irradiated from a drone, and is a collection of points on the surface of the bark around the trunk of a broad-leaved tree.

[0031] In the example shown in Figure 1, the trunk center point candidate calculation unit 15C calculates candidates for the center point of the trunk of a broad-leaved tree (a point on the center line of the trunk) based on the points contained in the trunk point cluster generated by the second clustering unit 15B (the points on the surface of the bark around the trunk of the broad-leaved tree). The trunk center point judgment unit 15D judges whether the candidate for the trunk center point calculated by the trunk center point candidate calculation unit 15C is the trunk center point based on the number of candidates for the trunk center point calculated by the trunk center point candidate calculation unit 15C. Specifically, when the number Kn of adjacent trunk center point candidates in the vertical direction among the trunk center point candidates calculated by the trunk center point candidate calculation unit 15C is greater than the threshold value Kc (i.e., when there are a sufficient number (density) of trunk center point candidates that can form the center line of the trunk), the trunk center point judgment unit 15D judges that the trunk center point candidate is the trunk center point. If the number Kn of adjacent trunk center point candidates in the vertical direction among the trunk center point candidates calculated by the trunk center point candidate calculation unit 15C is less than or equal to the threshold value Kc, the trunk center point judgment unit 15D judges that the trunk center point candidate is not a trunk center point but noise. In the example shown in Figure 8, the second clustering unit 15B generates a trunk point cloud cluster that extends throughout the entire height direction (up and down direction in Figure 5) of the corrected three-dimensional point cloud data, but in other examples, for example, the second clustering unit 15B may use a trunk center point that is determined to be a trunk center point by the trunk center point determination unit 15D to be a trunk center point, to generate a trunk center point cluster (clustered trunk center points) that extends throughout the entire height direction (up and down direction in Figure 5) of the corrected three-dimensional point cloud data.

[0032] As mentioned above, broadleaf trees have more trunk bending and branching than coniferous trees. Therefore, the trunk point cloud clusters stacked in the height direction as shown in Figure 8 may contain parts (branching parts) where it is difficult to distinguish whether a branch is branching from the trunk of a single broadleaf tree, or whether two broadleaf trees, including one with a large trunk bending, are close to each other.

[0033] Therefore, in the example shown in Figure 1, when a trunk point cloud cluster including a branching portion (a portion that may be a branch branching off from the trunk) is generated by the second clustering unit 15B, the separation processing unit 15E determines whether the trunk point cloud cluster including the branching portion corresponds to two different broad-leaved trees, or whether the trunk point cloud cluster including the branching portion corresponds to a single broad-leaved tree having a trunk and a branch branching off from the trunk. The separation processing unit 15E judges whether the trunk point cloud cluster including the branching part corresponds to two different broad-leaved trees or corresponds to one broad-leaved tree based on the height dimension of the branching part (part that may be a branch branched from the trunk). In other words, the separation processing unit 15E judges whether the trunk point cloud cluster including the branching part corresponds to two different broad-leaved trees or corresponds to one broad-leaved tree based on the number of slice data including trunk point cloud clusters that may correspond to branches branched from the trunk. In detail, the separation processing unit 15E judges that the trunk point cloud cluster including the branching part corresponds to two different broad-leaved trees when the height dimension of the branching part (part that may be a branch branched from the trunk) is two-thirds or more of the height dimension of the corrected three-dimensional point cloud data created by the three-dimensional point cloud data correction unit 13. In other words, the separation processing unit 15E judges that the trunk point cloud cluster including the branching part corresponds to two different broad-leaved trees, not to one broad-leaved tree having a trunk and a branch branched from the trunk, when the number of slice data including the trunk point cloud cluster that may correspond to a branch branched from the trunk is two-thirds or more of the total number of slice data (for example, 96 in the example shown in FIG. 5). On the other hand, the separation processor 15E determines that a trunk point cloud cluster including a branching part corresponds to one broadleaf tree when the height dimension of the branching part (a part that may be a branch branching from the trunk) is less than two-thirds of the height dimension of the corrected three-dimensional point cloud data created by the three-dimensional point cloud data corrector 13. In other words, the separation processor 15E determines that the branching part is highly likely to be a branch branching from the trunk when the number of slice data including trunk point cloud clusters that may correspond to a branch branching from the trunk is less than two-thirds of the total number of slice data (for example, 96 in the example shown in FIG. 5).

[0034] In the example shown in Figure 1, as described above, the separation processing unit 15E determines whether a trunk point cloud cluster including a branching portion corresponds to two different broad-leaved trees or whether the trunk point cloud cluster including a branching portion corresponds to one broad-leaved tree based on the number of slice data including the trunk point cloud cluster. In another example, the separation processing unit 15E may determine whether a trunk point group cluster including a branching portion corresponds to two different broad-leaved trees or whether a trunk point group cluster including a branching portion corresponds to one broad-leaved tree based on the number of slice data including a trunk center point cluster.

[0035] Fig. 9 is a diagram for explaining an example in which the separation processing unit 15E judges that the trunk point cloud cluster including the branching part corresponds to two different broad-leaved trees. In detail, Fig. 9(A) shows the state before the separation processing unit 15E judges whether the trunk point cloud cluster including the branching part corresponds to two different broad-leaved trees or whether the trunk point cloud cluster including the branching part corresponds to one broad-leaved tree. Fig. 9(B) shows the state after the separation processing unit 15E judges that the trunk point cloud cluster including the branching part corresponds to two different broad-leaved trees. In the example shown in Figure 9, the number of slice data including trunk center point clusters that may correspond to branches branching off from the trunk is more than two-thirds of the total number of slice data, so the separation processing unit 15E determines that the trunk point group cluster including the branching part (the part that appears to be a branch branching off from the trunk) does not correspond to one broad-leaved tree having a trunk and a branch branching off from the trunk, but rather corresponds to two different broad-leaved trees.

[0036] In the example shown in Figure 1, the individual tree judgment unit 15F judges, based on the point judged to be the trunk center point by the trunk center point judgment unit 15D and the result of the judgment by the separation processing unit 15E, whether the trunk point group cluster to be judged corresponds to each of the trunks of multiple broad-leaved trees included in the forest area to be surveyed, or is noise. In detail, the single tree judgment unit 15F uses, for example, the trunk center point cluster extending in the vertical direction generated by the second clustering unit 15B and the results of judgment by the separation processing unit 15E to judge whether the trunk point group cluster shown in Figure 8, for example, corresponds only to the trunks of multiple broad-leaved trees contained in the forest area surveyed, or whether it includes branches branching off from the trunks (which should be removed as noise).

[0037] FIG. 10 is a diagram showing an example of a trunk point cloud cluster determined by the single tree determination unit 15F to correspond to the trunks of multiple broad-leaved trees contained in the survey target forest area. In the example shown in FIG. 10, a serial number is assigned to each stem point cluster representing each of the multiple broad-leaved trees included in the surveyed forest area.

[0038] In the example shown in Figure 1, the single tree point cloud extraction unit 16 reproduces parts other than the trunks of multiple broad-leaved trees included in the surveyed forest area by using 3D point cloud data (for example, 3D point cloud data of parts corresponding to the branches and leaves of broad-leaved trees) from the 3D point cloud data (see Figures 2 and 3) created by the 3D point cloud data creation unit 12 other than the 3D point cloud data determined by the single tree judgment unit 15F to be the trunk point cloud cluster corresponding to the trunk (for example, the trunk point cloud cluster shown in Figure 10).

[0039] FIG. 11 is a diagram showing an example of 3D point cloud data of multiple broad-leaved trees included in the surveyed forest area after the single tree point cloud extraction unit 16 has reproduced parts other than the tree trunks. In the example shown in Figure 11, the tree shape (trunk curvature, branch structure) of each of the multiple broad-leaved trees contained in the surveyed forest area is reproduced, while each of the multiple broad-leaved trees contained in the surveyed forest area is clearly identified.

[0040] In the example shown in Figure 1, the individual tree height analysis unit 17 calculates the horizontal position (e.g., latitude, longitude, etc.) and tree height of each of the multiple broad-leaved trees included in the surveyed forest area by using the 3D point cloud data that has been determined by the individual tree judgment unit 15F to be a trunk point cloud cluster corresponding to the trunk, and the 3D point cloud data that has been used by the individual tree point cloud extraction unit 16 to reproduce parts other than the trunk.

[0041] FIG. 12 is a diagram showing data in which information indicating the horizontal position and tree height calculated by single tree height analysis unit 17 is added to the three-dimensional point cloud data shown in FIG.

[0042] Fig. 13 is a flowchart for explaining an example of processing executed in the forest resource information calculation device 1 of the first embodiment. Fig. 14 is a flowchart for explaining in detail an example of processing executed in step S50 of Fig. 13. In the example shown in Figures 13 and 14, in step S10, the laser measurement data acquisition unit 11 acquires laser measurement data obtained by using a drone to irradiate the forest area to be surveyed with laser light from multiple directions. Next, in step S20, the three-dimensional point cloud data creation unit 12 creates three-dimensional point cloud data based on the laser measurement data acquired in step S10. Next, in step S30, the 3D point cloud data correcting unit 13 creates corrected 3D point cloud data, which is 3D point cloud data that excludes the influence of the inclination of the earth's surface, from the 3D point cloud data created in step S20. Next, in step S40, the slice data generating unit 14 generates a plurality of slice data by dividing the corrected three-dimensional point cloud data generated in step S30 in the height direction.

[0043] Next, in step S50, the single tree analysis unit 15 analyzes each of the broad-leaved trees contained in the survey target forest area (see FIG. 2). In detail, in step S51, the single tree analysis unit 15 sets an initial value of the raw point cloud neighboring distance Dr, an initial value of the raw point cloud lowest neighboring point number Kr, an initial value of the center point neighboring distance Dc, and an initial value of the center point lowest neighboring point number Kc. Next, in step S52, the first clustering unit 15A performs clustering of the three-dimensional point cloud data contained in each of the multiple slice data created in step S40, thereby generating, for each slice data, a trunk point cloud cluster corresponding to each of the trunks of multiple broad-leaved trees contained in the surveyed forest area. In step S53, the first clustering unit 15A determines whether or not each of the three-dimensional point groups included in each slice data is a low-density point group. In step S53A, the first clustering unit 15A determines low-density points among the three-dimensional points included in each slice data as noise corresponding to twigs, leaves, and the like, and removes them from each slice data. In step S54, the second clustering unit 15B stacks the trunk point cloud clusters contained in each of the adjacent slice data throughout the entire height direction of the corrected 3D point cloud data (the up and down direction in Figure 5), thereby generating trunk point cloud clusters corresponding to each of the trunks of multiple broad-leaved trees contained in the surveyed forest area.

[0044] Next, in step S55, the trunk center point candidate calculation unit 15C calculates candidates for the center point of the trunk of the broad-leaved tree based on the points contained in the trunk point cluster generated in step S54 (the points on the surface of the bark around the trunk of the broad-leaved tree). Next, in step S56, the trunk center point candidate calculation unit 15C performs a process of connecting (performing clustering) the candidates for the center points of the broad-leaved tree trunks calculated in step S55 in the height direction (up and down direction in Figure 5) of the corrected 3D point cloud data. Next, in step S57, the trunk center point determination unit 15D determines whether the candidate for the center point of the trunk calculated in step S55 is the trunk center point, based on the number of candidates for the center point of the trunks connected in the height direction in step S56. Specifically, the trunk center point determination unit 15D determines whether the candidate for the center point of the trunks connected in the height direction in step S56 is distant from the candidates for the center point of other trunks (i.e., the trunks are not connected in the height direction), and whether the candidate for the center point of the trunks connected in the height direction in step S56 is deviated from the trunks. In step S57A, the trunk center point judgment unit 15D judges trunk center point candidates that are away from other trunk center point candidates and trunk center point candidates that are off the trunk as noise center points and removes them. In step S58, the trunk center point determination unit 15D determines that the candidate trunk center points that were not determined to be noise center points are trunk center points, and the second clustering unit 15B performs clustering of the trunk center points (connecting the trunk center points in the vertical direction) to generate trunk center point clusters corresponding to each of the multiple broad-leaved trees included in the surveyed forest area.

[0045] Next, in step S59, the separation processing unit 15E determines whether the trunk center point cluster including the branching portion (a portion that may be a branch branching off from the trunk) corresponds to two different broad-leaved trees, or whether the trunk point group cluster including the branching portion corresponds to a single broad-leaved tree having a trunk and a branch branching off from the trunk. In detail, the separation processing unit 15E judges that the trunk center point cluster including the branching part corresponds to two different broad-leaved trees when the height dimension of the branching part (part that may be a branch branched from the trunk) is two-thirds or more of the height dimension of the corrected three-dimensional point cloud data created in step S30. In other words, the separation processing unit 15E judges that the trunk center point cluster including the branching part corresponds to two different broad-leaved trees, not to one broad-leaved tree having a trunk and a branch branched from the trunk, when the number of slice data including the trunk center point cluster that may correspond to a branch branched from the trunk is two-thirds or more of the total number of slice data (for example, 96 in the example shown in FIG. 5). In step S59A, the separation processing unit 15E changes the value of the center-point neighboring distance Dc. In step S59B, the separation processing unit 15E determines that trunk center points having a center-point neighboring distance Dc value of 1 or less are noise center points, and removes them.

[0046] Next, in step S60, the individual tree judgment unit 15F judges, based on the point judged to be the trunk center point in step S57 and the result of the judgment in step S59, whether the trunk point group cluster being judged corresponds to each of the trunks of multiple broad-leaved trees included in the surveyed forest area, or whether it is noise. In step S60, the single tree judgment unit 15F uses the trunk center point cluster extending in the height direction generated in step S58 and the result of the judgment in step S59. The conditions under which the single tree judgment unit 15F determines that the trunk point cloud cluster being judged corresponds to each trunk of a plurality of broad-leaved trees contained in the surveyed forest area (i.e., the trunk point cloud cluster being judged corresponds to an individual tree) include, for example, (1) there are five or more trunk center point clusters generated in step S58 within a height range of 3m to 8m, (2) the minimum height of the trunk center point clusters generated in step S58 is 4m or less, (3) the maximum height of the trunk center point clusters generated in step S58 is 5m or more, (4) if the trunk center point clusters connected in the vertical direction include a branched portion (a portion that may be a branch branching off from the trunk), the number of slice data containing the branched portion is one-third or less of the total number of slice data (for example, 96 in the example shown in Figure 5), and (5) the trunk center point cluster generated in step S58 is included in 10 or more consecutive slice data in the vertical direction. If the trunk point cluster being judged does not correspond to the trunks of each of the multiple broad-leaved trees contained in the forest area being surveyed and is judged to be noise, in step S60A, the trunk center point cluster extending in the vertical direction generated in step S58 is judged to be a noise center point cluster and removed.

[0047] Next, in step S70, the single tree point cloud extraction unit 16 reproduces parts other than the trunks of multiple broad-leaved trees included in the surveyed forest area by using 3D point cloud data (for example, 3D point cloud data of parts corresponding to the branches and leaves of broad-leaved trees) from the 3D point cloud data created in step S20 (see Figures 2 and 3) other than the 3D point cloud data determined in step S60 to be the trunk point cloud cluster corresponding to the trunk (for example, the trunk point cloud cluster shown in Figure 10). Next, in step S80, the individual tree height analysis unit 17 calculates the horizontal position (e.g., latitude, longitude, etc.) and tree height of each of the multiple broad-leaved trees included in the surveyed forest area by using the 3D point cloud data determined to be the trunk point cloud cluster corresponding to the trunk in step S60 and the 3D point cloud data used to reproduce parts other than the trunk in step S70.

[0048] [Example] The inventors used an M600Pro manufactured by DJI as a drone (UAV: Unmanned Aerial Vehicle) and a Surveyor Ultra manufactured by YellowScan as a laser measurement device to perform laser measurements of designated survey target forest areas in Yashiro, Futatsuya, and Kijimadaira villages in Omachi city, and calculated forest resource information using the forest resource information calculation device 1 of the first embodiment.

[0049] Figure 15 is a diagram explaining the forest area surveyed in Shasha, Omachi city, the first research site. Figure 16 is a diagram explaining the forest area surveyed in Futatsuya, Omachi city, the second research site. Figure 17 is a diagram explaining the forest area surveyed in Kijimadaira village, the third research site. Figure 18 shows the results of field surveys at the three research sites. Fig. 19 is a diagram showing the classification results of the first and second survey sites. In detail, Fig. 19(A) shows the horizontal positions of multiple broad-leaved trees included in the surveyed forest area of ​​Omachi City, calculated by the single tree height analysis unit 17 of the forest resource information calculation device 1 of the first embodiment, as "ULS ITD (single tree extraction result)", and Fig. 19(B) shows the horizontal positions of multiple broad-leaved trees included in the surveyed forest area of ​​Futatsuya, Omachi City, calculated by the single tree height analysis unit 17 of the forest resource information calculation device 1 of the first embodiment, as "ULS ITD". Fig. 20 is a diagram showing the classification results of the third survey area, etc. In detail, Fig. 20(A) shows the horizontal positions of each of a plurality of broad-leaved trees included in the surveyed forest area of ​​Kijimadaira village, calculated by the single tree height analysis unit 17 of the forest resource information calculation device 1 of the first embodiment, in "ULS ITD", and Fig. 20(B) shows the accuracy rate of the calculation results by the single tree height analysis unit 17 of the forest resource information calculation device 1 of the first embodiment for the field survey results in the three survey areas, etc.

[0050] The accuracy rate in the surveyed area of ​​Omachi City was 100%. In other areas, the accuracy rate exceeded 90% for trees with DBH (diameter at breast height) exceeding 10 cm. Since the accuracy rate at a practical level is over 80%, it can be said that the practical accuracy of the forest resource information calculation device 1 of the first embodiment is guaranteed.

[0051] We attempted to extract individual tree classifications and crown shapes from 3D point cloud data in three different types of broadleaf forests in three locations. As a result, we found that the models could be represented in three dimensions, demonstrating that the first embodiment of the forest resource information calculation device 1 has high spatial reproduction capabilities. In different types of broadleaf forests dominated by chestnut, oak, and beech, the accuracy rate of individual tree classification was nearly 100%. Even in the diverse and complex Omachi City Shrine plot, where 15 species are mixed, the false positive rate was less than 3%, meaning there is no problem with practical accuracy.

[0052] As mentioned above, broadleaf forests are a mixture of many different tree species, have high tree density, have round crowns, and are difficult to find the tops of trees. This makes it difficult to classify individual trees using pattern recognition or AI (artificial intelligence). In addition, broadleaf trees have a lot of leaves, bent trunks, and branching, so it is difficult to distinguish individual trees using aerial images. Therefore, in the forest resource information calculation device 1 of the first embodiment, the above-mentioned processing is carried out. In other words, in order to digitize the trunks and branches from the base to the top, laser measurements are taken using a drone during the leaf-shedding season, obtaining high-density 3D point cloud data. In addition, three-dimensional point cloud data within a range is extracted at level slices spaced at regular intervals, and clustering is performed based on point density to generate trunk point cloud clusters for each slice data. Furthermore, by stacking the trunk point cloud clusters for each slice data in the vertical direction, a trunk point cloud cluster extending in the vertical direction is generated with noise such as branches and leaves removed, and the standing tree positions are extracted. In addition, a 3D model of the tree trunk up to the apex is created by extracting trunk diameter point clouds by hierarchy and connecting the trunk diameters.

[0053] According to an example of the first embodiment of the forest resource information calculation device 1 (results of research by the inventors et al.), it is possible to classify individual trees and grasp tree shapes in three-dimensional space in a wide-area broadleaf forest based on laser measurement data using only a drone, without the need for on-site surveys. In particular, according to the forest resource information calculation device 1 of the first embodiment, the position, height, and shape of trees are calculated, and the amount of timber output can be determined by accumulating the number of trees.

[0054] <Second embodiment> A second embodiment of the forest resource information calculation device, forest resource information calculation method, and program of the present invention will be described below. The forest resource information calculation device 1 of the second embodiment is configured similarly to the forest resource information calculation device 1 of the first embodiment described above, except for the points described below. Therefore, according to the forest resource information calculation device 1 of the second embodiment, it is possible to achieve the same effects as the forest resource information calculation device 1 of the first embodiment described above, except for the points described below.

[0055] In the forest resource information calculation device 1 of the first embodiment, the processes shown in FIG. 13 and FIG. 14 are executed by using one computer. On the other hand, in the forest resource information calculation device 1 of the second embodiment, the processes shown in Fig. 13 and Fig. 14 may be executed by using multiple computers. Specifically, for example, the processes of steps S10, S20, S30, etc. may be executed in, for example, a cloud server.

[0056] Although the present invention has been described above using the embodiments, the present invention is not limited to these embodiments, and various modifications and substitutions can be made without departing from the scope of the present invention. The configurations described in the above-mentioned embodiments and examples may be combined.

[0057] In addition, all or part of the functions of each unit of the forest resource information calculation device 1 in the above-mentioned embodiment may be realized by recording a program for realizing these functions on a computer-readable recording medium, and reading and executing the program recorded on the recording medium into a computer system. Note that the "computer system" referred to here includes hardware such as the OS and peripheral devices. In addition, the term "computer-readable recording medium" refers to portable media such as flexible disks, optical magnetic disks, ROMs, and CD-ROMs, and storage units such as hard disks built into computer systems. Furthermore, the term "computer-readable recording medium" may also include those that dynamically hold a program for a short period of time, such as a communication line when transmitting a program via a network such as the Internet or a communication line such as a telephone line, and those that hold a program for a certain period of time, such as a volatile memory inside a computer system that serves as a server or client in such cases. Furthermore, the above program may be one that realizes part of the above-mentioned functions, or may be one that can realize the above-mentioned functions in combination with a program already recorded in the computer system. [Explanation of symbols]

[0058] 1...Forest resource information calculation device, 11...Laser measurement data acquisition unit, 12...3D point cloud data creation unit, 13...3D point cloud data correction unit, 14...Slice data creation unit, 15...Single tree analysis unit, 15A...First clustering unit, 15B...Second clustering unit, 15C...Tree trunk center point candidate calculation unit, 15D...Tree trunk center point judgment unit, 15E...Separation processing unit, 15F...Single tree judgment unit, 16...Single tree point cloud extraction unit, 17...Single tree height analysis unit

Claims

1. a laser measurement data acquisition unit that acquires laser measurement data obtained by irradiating a survey target forest area with laser light from multiple directions; a three-dimensional point cloud data creation unit that creates three-dimensional point cloud data based on the laser measurement data acquired by the laser measurement data acquisition unit; a three-dimensional point cloud data correction unit that creates corrected three-dimensional point cloud data, which is three-dimensional point cloud data obtained by excluding the influence of the inclination of the earth's surface from the three-dimensional point cloud data created by the three-dimensional point cloud data creation unit; a slice data creation unit that creates a plurality of slice data by dividing the corrected three-dimensional point cloud data created by the three-dimensional point cloud data correction unit in a height direction; a single tree analysis unit that analyzes each of a plurality of broad-leaved trees included in the survey target forest area; The single tree analysis unit a first clustering unit that performs clustering of three-dimensional point cloud data included in each of the plurality of slice data created by the slice data creation unit to generate a trunk point cloud cluster corresponding to each trunk of a plurality of broad-leaved trees included in the survey target forest area for each slice data; a second clustering unit that generates a trunk point cloud cluster corresponding to each of the trunks of a plurality of broad-leaved trees included in the surveyed forest area by stacking a plurality of trunk point cloud clusters included in one slice data and a plurality of trunk point cloud clusters included in adjacent slice data, which is slice data adjacent to the one slice data, over the entire height direction of the corrected 3D point cloud data; a separation processing unit that, when a trunk point cloud cluster including a branching portion is generated by the second clustering unit, judges whether the trunk point cloud cluster including the branching portion corresponds to two different broad-leaved trees, or whether the trunk point cloud cluster including the branching portion corresponds to one broad-leaved tree having a trunk and a branch branching from the trunk. Forest resource information calculation device.

2. The separation processing unit determines whether the trunk point cloud cluster including the branching portion corresponds to two different broad-leaved trees or whether the trunk point cloud cluster including the branching portion corresponds to one broad-leaved tree based on the height dimension of the branching portion. The forest resource information calculation device according to claim 1.

3. The separation processing unit includes: if the height dimension of the branching portion is equal to or greater than two-thirds of the height dimension of the corrected three-dimensional point cloud data created by the three-dimensional point cloud data correction unit, it is determined that the trunk point cloud cluster including the branching portion corresponds to two different broad-leaved trees; if the height dimension of the branch portion is less than two-thirds of the height dimension of the corrected three-dimensional point cloud data created by the three-dimensional point cloud data correction unit, it is determined that the trunk point cloud cluster including the branch portion corresponds to one broad-leaved tree. The forest resource information calculation device according to claim 2.

4. The single tree analysis unit a trunk center point candidate calculation unit that calculates candidates for the center point of the trunk of a broad-leaved tree based on the points included in the trunk point cluster generated by the second clustering unit; The forest resource information calculation device according to claim 1.

5. The single tree analysis unit and a trunk center point determination unit that determines whether the candidate for the trunk center point calculated by the trunk center point candidate calculation unit is a trunk center point based on the number of candidates for the trunk center point calculated by the trunk center point candidate calculation unit. The forest resource information calculation device according to claim 4.

6. The single tree analysis unit and a single tree judgment unit that judges whether the target trunk point cluster corresponds to each trunk of a plurality of broad-leaved trees included in the survey target forest area or is noise, based on the points judged to be trunk center points by the trunk center judgment unit and the result of judgment by the separation processing unit. The forest resource information calculation device according to claim 5.

7. a single tree point cloud extraction unit that reproduces parts of a plurality of broad-leaved trees included in the surveyed forest area other than the trunks by using three-dimensional point cloud data other than the three-dimensional point cloud data determined by the single tree determination unit to be a trunk point cloud cluster corresponding to the trunks, among the three-dimensional point cloud data created by the three-dimensional point cloud data creation unit; The forest resource information calculation device according to claim 6.

8. a single tree height analysis unit that calculates the horizontal position and tree height of each of a plurality of broad-leaved trees included in the surveyed forest area by using the three-dimensional point cloud data that is determined by the single tree determination unit to be a trunk point cloud cluster corresponding to the tree trunk and the three-dimensional point cloud data that is used by the single tree point cloud extraction unit to reproduce parts other than the tree trunk; The forest resource information calculation device according to claim 7.

9. a laser measurement data acquisition step of acquiring laser measurement data obtained by irradiating a survey target forest area with laser light from multiple directions; a three-dimensional point cloud data creation step of creating three-dimensional point cloud data based on the laser measurement data acquired in the laser measurement data acquisition step; a three-dimensional point cloud data correction step of creating corrected three-dimensional point cloud data, which is three-dimensional point cloud data obtained by excluding the influence of the inclination of the earth's surface from the three-dimensional point cloud data created in the three-dimensional point cloud data creation step; a slice data creation step of creating a plurality of slice data by dividing the corrected three-dimensional point cloud data created in the three-dimensional point cloud data correction step in a height direction; and a single tree analysis step of analyzing each of a plurality of broad-leaved trees included in the surveyed forest area, The single tree analysis step includes: a first clustering step of clustering three-dimensional point cloud data included in each of the plurality of slice data created in the slice data creation step, for each slice data, to generate a trunk point cloud cluster corresponding to each trunk of a plurality of broad-leaved trees included in the survey target forest area; a second clustering step of generating trunk point cloud clusters corresponding to the trunks of each of the broad-leaved trees included in the surveyed forest area by stacking a plurality of trunk point cloud clusters included in one slice data with a plurality of trunk point cloud clusters included in adjacent slice data, which is slice data adjacent to the one slice data, over the entire height direction of the corrected 3D point cloud data; and a separation processing step of determining, when a trunk point cloud cluster including a branching portion is generated in the second clustering step, whether the trunk point cloud cluster including the branching portion corresponds to two different broad-leaved trees, or whether the trunk point cloud cluster including the branching portion corresponds to one broad-leaved tree having a trunk and a branch branching from the trunk. Forest resource information calculation methods.

10. On the computer, a laser measurement data acquisition step of acquiring laser measurement data obtained by irradiating a survey target forest area with laser light from multiple directions; a three-dimensional point cloud data creation step of creating three-dimensional point cloud data based on the laser measurement data acquired in the laser measurement data acquisition step; a three-dimensional point cloud data correction step of creating corrected three-dimensional point cloud data, which is three-dimensional point cloud data obtained by excluding the influence of the inclination of the earth's surface from the three-dimensional point cloud data created in the three-dimensional point cloud data creation step; a slice data creation step of creating a plurality of slice data by dividing the corrected three-dimensional point cloud data created in the three-dimensional point cloud data correction step in a height direction; and a single tree analysis step of analyzing each of a plurality of broad-leaved trees included in the surveyed forest area, The single tree analysis step includes: a first clustering step of generating, for each slice data, a trunk point cloud cluster corresponding to each trunk of a plurality of broad-leaved trees included in the survey target forest area by clustering the three-dimensional point cloud data included in each of the plurality of slice data generated in the slice data generation step; a second clustering step of generating trunk point cloud clusters corresponding to the trunks of each of the broad-leaved trees included in the surveyed forest area by stacking a plurality of trunk point cloud clusters included in one slice data with a plurality of trunk point cloud clusters included in adjacent slice data, which is slice data adjacent to the one slice data, over the entire height direction of the corrected 3D point cloud data; and a separation processing step of determining, when a trunk point cloud cluster including a branching portion is generated in the second clustering step, whether the trunk point cloud cluster including the branching portion corresponds to two different broad-leaved trees, or whether the trunk point cloud cluster including the branching portion corresponds to one broad-leaved tree having a trunk and a branch branching from the trunk. program.

Images