JP7157434B2 - Forest resource information calculation method and forest resource information calculation device - Google Patents

Description

The present invention relates to a forest resource information calculation method and a forest resource information calculation apparatus for creating forest resource information related to a survey target forest area based on image data corresponding to an area including the survey target forest area.

Information on forest resources including tree height (referred to as forest resource information) is important basic information in forest management. Conventionally, forest resource information is obtained by conducting ground surveys, where surveyors set a small survey area in what is considered to be a standard forest, and measure the position, breast height diameter, and tree height of trees in the survey area. , by multiplying the result by the managed area, the method of estimating the overall forest resource information is widely used.

In order to improve the estimation accuracy of forest resources in such forest resource information estimation methods, it is necessary to increase the number of field survey locations, which requires a great deal of labor and cost. Tree height calculation, which is the most prone to errors in ground surveys, is often difficult to see the tops of trees from the ground due to obstacles and thick branches and leaves in the forest. is used to measure the tree height, and a regression equation is created from the relationship between the measured diameter at breast height and the height of the sample tree, and the tree height is estimated and interpolated. Furthermore, measurement error increases when affected by the skill of the measurer and the terrain such as the slope of the measurement.

Therefore, recently, tree vertices and tree extraction are also evaluated from three-dimensional point cloud data obtained by airborne laser measurement. In addition, images of forests captured from above by drones, aircraft, etc. are converted into 3D point cloud data to evaluate tree vertices and tree extraction.

Japanese Patent No. 4946072 Japanese Patent No. 4279894 Japanese Patent No. 5507418

The techniques described in Patent Literature 1 and Patent Literature 2 extract tree vertices and tree positions from three-dimensional point cloud data using a spatial filtering method called a known local maximum value filter, and extract them to a fixed size (3 × 3, 5 ×5, etc.) is moved on the tree crown high point group data, and the tree vertices with the maximum value in the region are collectively extracted.

The techniques described in Patent Documents 1 and 2 are effective methods for planted forests partitioned like a checker board and for simple and sparse coniferous trees, but most forests have different standing tree densities. ``There are many unextracted or erroneous extractions'', ``Depending on the size of the filter, it is not possible to extract both small and large crowns at the same time'', In high forests and clusters of trees, only one tree height can be extracted and there are many unextracted trees," and "In the gaps in the canopy of the forest, the underlying vegetation is mistakenly extracted as the top of the tree." there are many issues.

On the other hand, the technique described in Patent Document 3 is a method of detecting the tree position by planarly projecting the three-dimensional point cloud data of the forest acquired by airborne laser measurement. In addition, there is a problem that it is difficult to classify forests in which the size of the canopy is mixed for each single tree.

As described above, in the above Patent Documents 1 to 3, in a forest where the difference in tree density and the size of the tree height and the size of the crown are mixed, the height of each tree, the diameter at breast height, and the volume obtained from the height and diameter at breast height of each tree. There is a problem that forest resource information cannot be calculated with high accuracy. In this specification, the term "tree crown" refers to a portion formed by branches and leaves of each tree that constitutes a forest when viewed from above.

Therefore, the present invention was made to solve the above problems, and a forest resource information calculation method that can calculate forest resource information with high accuracy even in a forest where the difference in tree density, tree height, and crown size are mixed. It aims at providing a forest resource calculation information setting device.

[1] A forest resource information calculation method of the present invention is a forest resource information calculation method for creating forest resource information related to a survey target forest area, and includes three-dimensional point cloud data corresponding to an area including the survey target forest area. a survey target forest area image data creation processing step of creating survey target forest area image data based on geographical information; and a canopy of each tree existing in the survey target forest area based on the survey target forest area image data. A crown height image data creation processing step of creating crown height image data representing the height of the tree as the crown height, and among the tree crown heights obtained from the crown height image data, the layer having the maximum crown height is defined as the first layer. , when a plurality of layers having a predetermined layer width starting from the maximum crown height are set from the first layer toward the ground surface, a tree crown having a tree vertex in each layer including the first layer; and a hierarchical tree vertex extraction processing step of extracting tree vertices for each of the layers.

As described above, in the forest resource information calculation method of the present invention, based on the survey target forest area image data created based on the three-dimensional point cloud data corresponding to the area including the survey target forest area and the geographical information, , create crown height image data that represents the crown height of trees existing in the survey target forest area, and among the tree crown heights obtained from the crown height image data, the layer with the maximum crown height is set as the first layer, The tree vertices of the tree crown having tree vertices in each layer including the first layer are extracted for each layer.

As a result, according to the forest resource information calculation method of the present invention, it is possible to extract the tree crown corresponding to each tree with high precision regardless of the height of the trees to be surveyed existing in the forest area to be surveyed. Forest resource information can be calculated with high accuracy even in forests with mixed densities, tree heights, and crown sizes.

[2] In the forest resource information calculation method of the present invention, the crown height image data creation processing step generates meshed digital surface model data and digital elevation model data based on the survey target forest area image data. It is preferable to create tree crown height image data in which the tree crown height is obtained for each mesh by calculating the difference between the digital surface layer model data and the digital elevation model data for each mesh.

As a result, the tree crown height (the height of the tree crown from the ground surface) can be calculated with high accuracy for each mesh. In this case, the digital surface layer model data refers to data representing the altitude of the surface layer of trees and the like, and the digital altitude model data refers to data representing the altitude of the ground surface. Therefore, by taking the difference between the digital surface layer model data and the digital elevation model data, the tree crown height (the height of the tree crown from the ground surface) can be obtained for each mesh.

[3] In the forest resource information calculation method of the present invention, in the hierarchical tree vertex extraction processing step, using the crown height image data obtained by obtaining the crown height for each mesh, each tree including the first hierarchy A hierarchical crown high point cloud data corresponding to each tree crown existing in the hierarchy is created by a hierarchical crown high point cloud data creation processing step for creating the hierarchical crown high point cloud data creation step and the hierarchical crown high point cloud data creation processing step. The maximum value of the hierarchical crown high point group data corresponding to the individual crown is calculated for each hierarchy from the hierarchical crown high point group data corresponding to the individual crown, and the calculated maximum value is and a tree vertex calculation processing step for determining the vertex of the tree crown corresponding to the hierarchical tree crown high point group data.

As described above, in the forest resource information calculation method of the present invention, the tree crown high point group data for each layer corresponding to each tree crown existing in each layer including the first layer is created for each layer, and created for each layer. The maximum value of the hierarchical crown high point cloud data corresponding to each crown is calculated from the hierarchical crown high point cloud data corresponding to each crown. Here, hierarchical crown high point cloud data is created for each layer, and from the hierarchical crown high point cloud data corresponding to each tree crown created for each layer, the hierarchical crown high point cloud corresponding to each tree crown is obtained. In the process of calculating the maximum value of the data, for example, each tree crown is cut into slices for each layer like a CT (Computed Tomography) scan to create tree crown high point cloud data for each layer. It is a process of calculating the maximum value from the tree crown high point group data for each hierarchy obtained. By performing such processing, the maximum value of individual tree crowns existing in each layer can be calculated for each layer. Note that the “tree top of tree crown” calculated for each layer is the tree top of the tree having the tree crown, and the tree height of the tree can be obtained from the calculated tree top.

[4] In the forest resource information calculation method of the present invention, the tree vertex calculation processing step includes, when calculating the maximum value, among the hierarchical tree crown high point group data corresponding to the individual tree crowns, It is preferable to calculate the maximum value by excluding hierarchical crown high point group data corresponding to the same crown as the tree crown for which the hierarchical crown high point group data has already been created.

As a result, it is possible to calculate the maximum value of only the tree crown having a tree vertex in the current processing target layer among the layers. For example, if the current processing target layer is the second layer, the maximum value of the tree crown where the tree vertex exists in the layer above the second layer (first layer) may be incorrectly calculated as the tree vertex of the layer. can be prevented. Thereby, it is possible to calculate the maximum value only for the tree crown in which the tree vertex exists in the current processing target layer (for example, the second layer).

[5] In the forest resource information calculation method of the present invention, in the hierarchical tree vertex extraction processing step, each mesh of the tree crown height image data is set as a processing target mesh, and an averaging filter is superimposed on the processing target mesh. In addition, noise-removed tree crown height image data is created by sequentially performing processing for each processing target mesh to set the average value of the tree crown height of each mesh included in the averaging filter to the tree crown height of the processing target mesh. and the hierarchical crown height point group data creation processing step converts the noise-removed tree crown height image data created by the averaging step into crown height image data for each mesh. is used as the tree crown height image data obtained to create the hierarchical tree crown high point group data.

By performing such an averaging process, for example, if there is extreme unevenness in the crown (parts where branches and leaves are sparse compared to the surroundings, long branches and leaves, etc.), this can be removed as noise. can be used to create noise-removed canopy height image data. Then, in the hierarchical crown high point cloud data creation processing step, the noise-removed crown height image data is used as tree crown height image data to create hierarchical crown high point cloud data, thereby obtaining highly accurate hierarchical crown height data. Point cloud data can be created. The elongated branches and leaves refer to a portion of the tree crown where the branches and leaves protrude from the surrounding branches and leaves.

[6] In the forest resource information calculation method of the present invention, it is preferable that the floor width is set to a length in the range of 0.2m to 2.0m.

The layer width is not particularly limited, but if it is set too finely (short length), the number of layers increases, the number of processing increases, the amount of data increases, and the problem arises that the calculation takes time. On the other hand, if it is set too large (too long), there arises a problem that the accuracy of the tree crown high point group data for each hierarchy decreases. For this reason, it is preferable to set the floor width in the range of 0.2 m to 2.0 m. It is preferable to consider the tree height of trees existing in the area.

[7] In the forest resource information calculation method of the present invention, the three-dimensional point cloud data is obtained by overlapping and shifting photographed image data from the sky, and the three-dimensional shape restoration technology is preferably created by SFM (Structure from Motion).

By using SFM, which is a three-dimensional shape restoration technology, it is possible to create a map-corrected orthoimage without distortion from two-dimensional photographed image data. Then, based on the orthorectified image created in this way, three-dimensional point cloud data representing the height can be created.

[8] In the forest resource information calculation method of the present invention, the three-dimensional point cloud data may be created based on laser measurement data obtained by irradiating laser light from above.

In this way, three-dimensional point group data can also be created from laser measurement data obtained by irradiating a laser beam from above.

[9] In the forest resource information calculation method of the present invention, based on the tree vertices of the tree crowns extracted for each layer, area division is performed to divide the adjacent tree crowns from the crown height of the tree crowns into areas. It is preferable to further include a detailed crown image data creation processing step of creating fine crown image data in which individual crowns are segmented using an algorithm.

By using such an area division algorithm, it is possible to create precise crown image data (precise crown image data) in which adjacent crowns are divided into individual crowns. The precise canopy image data created in this manner is precise canopy image data that enables highly accurate extraction of individual canopies corresponding to the trees to be surveyed. For this reason, the precise crown information created based on the precise crown image data becomes highly accurate crown information corresponding to each tree of the survey target trees, and information on forest resources (forest resource information ), it is possible to obtain highly accurate forest resource information.

[10] In the forest resource information calculation method of the present invention, the area division algorithm is preferably the Watershed algorithm.

The Watershed Algorithm is an algorithm capable of classifying individual objects by extracting the boundaries of individual objects with high accuracy when a large number of closely spaced adjacent objects exist in a connected state. Therefore, by using the Watershed algorithm, densely existing tree crowns can be classified as individual tree crowns with high accuracy regardless of the size of the tree crown (crown diameter and height).

[11] In the forest resource information calculation method of the present invention, precise crown information creation processing for creating precise crown information corresponding to each tree from the precise crown image data created by the precise crown image data creation processing step. and a forest resource information calculation processing step of creating forest resource information about the survey target forest area based on the precise crown information corresponding to each tree created by the precise crown information creation processing step. and the precise crown information corresponding to each tree includes at least the label number of the crown corresponding to each tree, the crown position of the crown corresponding to each tree, the crown diameter of the crown corresponding to each tree, and the The crown area of the tree crown corresponding to each tree is included, and the forest resource information calculation processing step includes calculating the diameter at chest height of each tree by calculating the diameter at approximately the height of a person's chest as the diameter at chest height of each tree. a processing step, a wood volume calculation processing step of calculating the wood volume of each tree, the diameter at breast height of each tree, the wood volume of each tree, the precise crown information created in the precise crown information creation processing step, and the stratification and a forest resource information aggregation processing step of performing a process of aggregating the tree height of each tree obtained based on the tree vertices extracted in the tree apex extraction step.

The precise crown information created based on the precise crown image data is highly accurate crown information corresponding to each tree of the trees to be surveyed. For this reason, the forest resource information calculation processing step obtains highly accurate forest resource information by calculating information on forest resources (forest resource information) based on the precise canopy information created based on the precise canopy image data. be able to. As a result, it is possible to provide high-precision forest resource information with little measurement error and high objectivity for the entire forest area to be surveyed, for each subdivision, and for any range, so it can be used for forest management such as planning forest operations such as thinning. . Specifically, for example, the calculation result calculated by the forest resource information calculation processing step is registered in the attribute database as forest resource summary information indicating the summary of the forest resource, and the forest resource summary registered in the attribute database is registered in the attribute database. By displaying the information as a list on the display, it is possible to grasp the overview of the forest resources in the surveyed forest area.

[12] A forest resource information calculation device of the present invention is a forest resource information calculation device that creates forest resource information related to a survey target forest area, and includes three-dimensional point cloud data corresponding to an area including the survey target forest area. a survey target forest area image data creation unit that creates the survey target forest area image data based on the geographical information; A tree crown height image data creation unit that creates tree crown height image data representing the height of a tree crown as a tree crown height; , when a plurality of layers having a predetermined layer width starting from the maximum crown height are set from the first layer toward the ground surface, a tree crown having a tree vertex in each layer including the first layer; and a hierarchical tree vertex extracting unit that extracts the tree vertex for each of the layers.

[13] In the forest resource information calculation apparatus of the present invention, the crown height image data creation unit creates meshed digital surface model data and digital elevation model data based on the survey target forest area image data. Then, it is preferable to create crown height image data in which the crown height is obtained for each mesh by taking the difference between the digital surface layer model data and the digital elevation model data for each mesh.

[14] In the forest resource information calculation device of the present invention, the hierarchical tree vertex extraction unit includes:
Using the tree crown height image data obtained by obtaining the tree crown height for each mesh, layered tree crown high point group data corresponding to individual tree crowns existing in each layer including the first layer is created for each layer. and from the hierarchical crown high point cloud data corresponding to the individual tree crowns created by the hierarchical crown high point cloud data creation unit, a tree vertex calculation unit that calculates the maximum value of the tree crown high point group data for each layer, and uses the calculated maximum value as the tree vertex of the tree crown corresponding to the tree crown high point group data for each layer. preferably.

[15] In the forest resource information calculation device of the present invention, when calculating the maximum value, the tree vertex calculation unit, out of the tree crown high point group data by hierarchy corresponding to the individual tree crowns, It is preferable to calculate the maximum value while excluding hierarchical crown high point group data corresponding to the same crown as the tree crown for which hierarchical crown high point group data has already been created.

[16] In the forest resource information calculation device of the present invention, the hierarchical tree vertex extraction unit includes:
Each mesh of the tree crown height image data is set as a mesh to be processed, an averaging filter is superimposed on the mesh to be processed, and the average value of the tree crown heights of each mesh included in the averaging filter is the crown height of the mesh to be processed. is further included in the averaging processing unit for creating noise-removed tree canopy height image data by sequentially performing the processing of It is preferable to create the tree crown height point group data by hierarchy using the noise-removed tree crown height image data created by the averaging processing unit as the tree crown height image data obtained by obtaining the tree crown height for each mesh.

[17] In the forest resource information calculation device of the present invention, it is preferable that the layer width is set to a length in the range of 0.2m to 2.0m.

[18] In the forest resource information calculation device of the present invention, the three-dimensional point cloud data is obtained by overlapping and shifting photographed image data from the sky, and the three-dimensional shape restoration technology is preferably created by SFM (Structure from Motion).

[19] In the forest resource information calculation device of the present invention, the three-dimensional point group data may be created based on laser measurement data obtained by irradiating laser light from above.

[20] In the forest resource information calculation device of the present invention, based on the tree vertices of the tree crowns extracted for each of the layers, area division is performed to divide adjacent tree crowns into areas based on the crown height of the tree crowns. It is preferable to further have a precise canopy image data generating unit that generates precise canopy image data in which individual canopies are segmented using an algorithm.

[21] In the forest resource information calculation device of the present invention, the area division algorithm is preferably the Watershed algorithm.

[22] In the forest resource information calculation apparatus of the present invention, a precise crown information creation unit that creates precise crown information corresponding to each tree from the precise crown image data created by the precise crown image data creation unit; a forest resource information calculation unit that creates forest resource information about the survey target forest area based on the precise crown information corresponding to each tree created by the precise crown information creation unit; The precise tree crown information corresponding to the trees includes at least the crown label number corresponding to each tree, the crown position of the crown corresponding to each tree, the crown diameter of the crown corresponding to each tree, and the tree crown corresponding to each tree. The forest resource information calculation unit includes a chest-height diameter calculation unit that calculates the diameter at about the height of a person's chest in each tree as the chest-height diameter of each tree; A timber quantity calculation unit that calculates the timber volume of a tree, the diameter at breast height of each tree, the timber volume of each tree, the precise crown information created by the precise crown information creation unit, and the tree vertex extraction unit for each hierarchy extracted by the and a forest resource information totalizing unit that performs processing for totaling the tree height of each tree obtained based on the tree vertices.

The forest resource information calculation device of the present invention described in [12] to [22] also has the same effect as the corresponding forest resource information calculation method of the present invention described in [1] to [11]. effect is obtained.

It is a flowchart shown in order to demonstrate the forest resource information calculation method which concerns on embodiment. FIG. 4 is a diagram showing a survey target forest area image displayed on the display; It is the crown height image displayed on the display. FIG. 2 is a flow chart for explaining in detail the hierarchical tree vertex extraction process (step S50) shown in FIG. 1; FIG. FIG. 10 is a diagram showing first layer tree crown high point group areas A1, A2, A3, . . . displayed on the display; 6 is a diagram showing tree vertices a1, a2, a3, . . . extracted in the first layer tree crown high point group areas A1, A2, A3, . FIG. 10 is a diagram showing second layer tree crown high point group areas B1, B2, B3, . . . displayed on the display; FIG. 10 is a diagram showing third layer tree crown high point group areas C1, C2, C3, . . . displayed on the display; FIG. 10 is a diagram showing tree crown high point group areas for each layer from the first layer to the fifth layer and tree vertices in the tree crown high point group areas for each layer; Tree crown high point cloud data for each layer (step S53) and tree vertex calculation processing (step S54) are sequentially performed from the first layer (top layer) to the bottom layer, and the tree crown high point group for each layer is obtained. FIG. 4 is a diagram showing regions and tree vertices; FIG. 11 shows a detailed canopy image displayed on the display; FIG. 10 is a diagram showing a list of single tree resource information displayed on the display; FIG. 10 is a diagram showing a list of forest resource summary information displayed on the display; BRIEF DESCRIPTION OF THE DRAWINGS It is a figure shown in order to demonstrate the forest resource information calculation apparatus 1 which concerns on embodiment. 15 is a diagram showing the configuration of a hierarchical tree vertex extraction unit 50 shown in FIG. 14; FIG.

FIG. 1 is a flow chart for explaining a forest resource information calculation method according to an embodiment. A forest resource information calculation method according to an embodiment is a forest resource information calculation method for creating forest resource information about a survey target forest area based on image data corresponding to an area including the survey target forest area. Here, the image data corresponding to the area including the survey target forest area is acquired by taking an aerial photograph of the area including the survey target forest area from above. For aerial photography, an aircraft, a drone, or the like can be used, but in the forest resource information calculation method according to the embodiment, a popular drone is used.

In addition, in the forest resource information calculation method according to the embodiment, the trees to be surveyed refer to tall planted trees such as larch, Japanese red pine, Japanese cedar, and cypress, and the forest area to be surveyed refers to these trees. is a forest area in which a forest formed by trees has a predetermined size. In addition, when the "survey target tree" is explained as an individual tree, it may be simply abbreviated as "each tree". This also applies to the forest resource information calculation device according to the embodiment described later.

As shown in FIG. 1, the processing performed in the forest resource information calculation method according to the embodiment includes data input processing (step S10) for inputting aerial image data captured by a drone into an information processing device such as a personal computer; A 3D point cloud data creation process (step S20) for creating 3D point cloud data based on the input aerial image data, and a 3D point cloud created by the 3D point cloud data creation process (step S20) Survey target forest area image data creation processing (step S30) for creating survey target forest area image data based on the data and geographical information obtained from the geographic information system (GIS) 100; Based on the survey target forest area image data created by the image data creation process (step S30), tree crown height image data representing the crown height of each tree existing in the survey target forest area is created. Among the tree crown heights obtained from the tree crown height image data created by the height image data creation process (step S40) and the tree crown height image data creation process (step S40), the layer having the maximum crown height is the first layer (the highest layer). (upper layer), when a plurality of layers having a layer width of a predetermined length are set from the first layer toward the ground surface, the tree vertices of the tree crown where the tree vertices exist in each layer including the first layer are and tree vertex extraction processing for each hierarchy (step S50).

In addition to the processing of steps S10 to S50, precise crown image data creation processing (step S60) for creating fine crown image data in which individual tree crowns are divided, and precise crown image data creation processing (step S60). Precise crown information creation processing (step S70) for creating precise crown information corresponding to each tree from the precise crown image data created by the above, and corresponding to each tree created by the precise crown information creation processing (step S70) and forest resource information calculation processing (step S80) for creating forest resource information related to the survey target forest area based on the detailed canopy information.
In addition, the precise crown image data creation processing is based on the tree vertices of the tree crowns extracted for each layer by the layer-specific tree vertex extraction processing (step S50), and based on the tree crown height of the tree crown, the adjacent tree crowns. Using an area segmentation algorithm that divides the , create precise canopy image data in which individual canopies are segmented.

Further, the forest resource information calculation process (step S80) includes a chest-height diameter calculation process (step S81) for calculating the diameter at approximately the height of a person's chest in each tree as the chest-height diameter of each tree, and the calculation of the timber volume of each tree. Calculated timber volume calculation processing (step S82), breast height diameter of each tree, timber volume of each tree, precise crown information created in the precise crown information creation processing (step S70), tree vertex extraction processing for each hierarchy (step S50) and a forest resource information tallying process (step S83) for tallying the tree height of each tree obtained based on the extracted tree vertices.

Further, the hierarchical tree vertex extraction process (step S50) includes a plurality of processes, and each of these processes will be described in detail with reference to FIG.

Next, each process after the three-dimensional point cloud data creation process (step S20) among the above-described steps S10 to S80 will be described in detail. In the 3D point cloud data creation process (step S20), a predetermined area including the survey target area is photographed by a camera mounted on a drone while overlapping and shifting the photographed locations by, for example, about 80%, and the image is From the data (aerial image data), SFM (Structure from Motion), which is a well-known 3D shape restoration technology, creates a map-corrected orthoimage without distortion, and 3D point cloud data from the created orthoimage. to create

The total area of the predetermined area including the survey target forest area is about 2 to 3 hectares. With a camera mounted on a drone, several hundred images of the entire predetermined area, including the survey target forest area with such a large area, are taken while overlapping about 80% and shifting, and from the captured images, SFM produces map-corrected orthoimages without distortion. Three-dimensional point cloud data representing height can be created based on the orthoimage created in this way. The SFM is described in the following publication 1.

Publication 1: Linda G. Shapiro; George C. Stockman (2001). Computer Vision. Prentice Hall.

The survey target forest area image data creation process (step S30) is based on geographic information obtained from the geographic information system (GIS) 100 (in this case, forests registered in forest management ledgers maintained by prefectures, etc.). The image data of the forest area to be surveyed (surveyed forest area) is extracted from the three-dimensional point cloud data using the boundary data of the area), and is used as the surveyed forest area image data. Then, the survey target forest area image corresponding to the survey target forest area image data is displayed on the display.

FIG. 2 is a diagram showing a survey target forest area image displayed on the display. FIG. 2 shows an extracted part of the survey target forest area. According to the survey target forest area image shown in FIG. 2, the crown of each tree can be identified. Since FIG. 2 is a monochrome image, it is difficult to identify the crown of each tree, but the crown of each tree can be easily identified on the color image that is the basis of FIG.

After that, based on the survey target forest area image data, crown height image data creation processing (step S40) is performed to create crown height image data representing the crown height of trees existing in the survey target forest area. The crown height image data creation process (step S40) calculates the crown heights of trees existing in the survey target forest area based on the survey target forest area image data created by the survey target forest area image data creation process (step S20). Create crown height image data representing (the height of the crown from the ground surface).

Specifically, meshed digital surface layer model data (data representing the elevation of the surface layer of trees, etc.) and digital elevation model data (data representing the elevation of the ground surface) are created based on the image data of the surveyed forest area. , the difference between the digital surface layer model data and the digital elevation model data is taken for each mesh to create tree crown height image data representing the tree crown height for each mesh. A tree crown height image corresponding to the tree crown height image data thus created is displayed on the display.

The mesh size can be set based on the ground resolution of the forest area image to be surveyed, but if the mesh size is too large, problems may arise in the accuracy of the calculated forest resource information. Conversely, if the mesh size is too small, the amount of data to be processed increases. Therefore, it is preferable to set the mesh size to an optimum size in consideration of the area of the survey target forest area, the state of the forest in the survey target forest area, the accuracy of the calculated forest resource information, and the like. In an embodiment, the mesh size is 10 cm x 10 cm.

FIG. 3 is a canopy height image displayed on the display. Note that the gradation from black to white in the canopy height image shown in FIG. 3 is the change in height with respect to the ground surface (0.0 m) as the reference (for example, 0.0 m is black). represents. At this stage, the tree crown height image data corresponding to the tree crown height image shown in FIG. 3 does not have a clear crown corresponding to each tree.

Then, based on the tree crown height image data created by the tree crown height image data creation process (step S40), the hierarchical tree vertex extraction process (step S50) is performed.

FIG. 4 is a flowchart for explaining in detail the hierarchical tree vertex extraction process (step S50) shown in FIG. As shown in FIG. 4, the hierarchical tree vertex extraction process (step S50) includes an averaging process (step S51), an initial value setting process (step S52), and a hierarchical crown high point group data creation process (step S53). , a tree vertex calculation process (step S54). In addition to these processes, calculation processes such as Hmax−D×Ni=Hi (step S55), Hi>0 (step S56), and Ni=Ni+1 (step S57) are included. Note that Hmax is the maximum crown height in the survey target forest area, D is the layer width, Ni is the current processing target layer, and Hi is the maximum tree crown term in the next processing target layer, which will be described later.

The averaging process (step S51) is performed when, for example, the crown height image shown in FIG. , are removed as noise to create tree crown height image data from which the noise has been removed. In the averaging process (step S51), each mesh of the tree canopy height image data created by the tree canopy height image data creating process (step S40) is set as a processing target mesh, and an averaging filter is superimposed on the processing target mesh. , the average value of the tree crown height of each mesh included in the averaging filter is set as the tree crown height of the processing target mesh. By sequentially performing such processing on each mesh to be processed, noise-removed canopy height image data is created.

The averaging process will be simplified and specifically explained. When the size of one mesh is assumed to be 10 cm on one side, the averaging filter (filter size is, for example, 5 meshes × 5 meshes) are superimposed, and the average value of the crown heights of the meshes included in the averaging filter is used as the crown height of the processing target mesh.

For example, the crown height of each mesh included in the range superimposed on the averaging filter (referred to as the averaging processing range) (for example, 5 meshes x 5 meshes, totaling 25 meshes) is integrated, and the The value obtained by dividing the total value by 25 is taken as the average value of the averaging processing range, and the average value is taken as the crown height of the processing target mesh. Such processing is performed for all meshes included in the canopy height image data. Note that the size of the averaging filter can be set based on, for example, the average crown diameter of survey target trees existing in the survey target forest area.

When such averaging is performed, the crown height for each mesh obtained by averaging may differ from the original crown height (crown height for each mesh). If the size is as small as 10 cm per side, the crown height for each mesh obtained by the averaging process can obtain a crown height that is not significantly different from the original crown height. In addition, by performing such an averaging process, extreme unevenness in individual tree crowns (parts where branches and leaves are sparse compared to the surroundings, elongated branches and leaves, etc.) are removed, and individual tree crowns are smoothed. Therefore, highly accurate segmentation can be performed when segmentation is performed by the Watershed algorithm, which will be described later.

The initial value setting process (step S52) is a process of setting an initial value when performing the hierarchical tree crown high point group data creation process (step S53). ) of the tree crown heights for each mesh (the maximum tree crown height in the survey target forest area) Hmax, and when the hierarchy is set downward (to the ground) from the maximum tree crown height Hmax as the starting point , and the current processing target hierarchy Ni.

Note that the maximum crown height Hmax among the tree crown heights of each mesh means that, among the tree crown heights obtained in each mesh, the crown height obtained in a certain mesh is, for example, 25 m, and the crown height of 25 m is Assuming that it is the maximum crown height among the tree crown heights obtained for all meshes, Hmax=25 m. The story width D is set to D=1 when the story width is set downward (to the ground) from the maximum tree crown height Hmax, for example, every 1 m. Also, the processing target layer Ni is set to Ni=1 because the layer to be processed in the initial stage is the first layer (top layer).

Thus, Hmax=25, D=1, and Ni=1 are set as the initial values, and the tree crown high point group data creation process for each layer (step S53) is performed. In this case, since Ni=1, the tree crown high point group data creation process for each layer in the first layer (uppermost layer) is performed. Since Hmax=25m and D=1, the first layer is 25m to 24m.

That is, the hierarchical tree crown high point group data creation process (step S53) selects the layer having the maximum tree crown height (Hmax=25 m) among the tree crown heights of each mesh calculated by the averaging process as the first layer (maximum As the upper layer), a layer width (D = 1 m) of a predetermined length is set starting from the maximum tree crown height (Hmax), and from the maximum tree crown height Hmax = 25 m, each layer width exists in each layer including the first layer. Hierarchical tree crown high point group data corresponding to individual tree crowns are created sequentially for each of the hierarchies.

The tree crown high point group data creation process for each layer (step S53) will be specifically described. In this case, the maximum crown height Hmax among the tree crown heights of each mesh is 25m, and the layer width is 1m, so the initial values are set to Hmax=25 and Ni=1 (first layer). . In this case, first, tree crown height point group data (referred to as first layer crown high point group data) corresponding to tree crowns belonging to the first layer (24 m to 25 m) among tree crown heights of each mesh is created. The first floor is 24m to 25m, but strictly speaking, it is 24.01m to 25m, from a value slightly exceeding 24m to 25m. This is the same for the layers lower than the first layer. do.
Then, the first layer crown high point group area corresponding to the first layer crown high point group data created by the hierarchical tree crown high point group data creating process (step S53) is displayed on the display.

Here, the first layer is set to 24.01 m to 25 m, and the second layer is set to 23.01 m to 24 m. .99 m, and the second layer may be set to 23 m to 23.99 m. This also applies to other hierarchies. However, as for the first layer, since the first layer is the top layer and there is no layer above it, the height may be 24 m to 25 m.

FIG. 5 is a diagram showing the first layer crown high point group area displayed on the display. In FIG. 5, the black-filled areas are the first layer crown high point group areas A1, A2, A3, . indicates the presence of a tree whose vertex reaches the first layer (24m to 25m). For example, in FIG. 5, looking at the state (cross-sectional view) of the group of trees contained in the black frame, among these groups of trees, the vicinity of the tree apex reaches the first layer indicated by a band. It can be seen that there are trees with

In FIG. 5, there are actually a large number of first layer crown high point group areas A1, A2, A3, . . . It is The same applies to tree crown high point group regions extracted in other layers, which will be described later.

Further, since FIG. 5 is a monochrome image, in FIG. 5, the first layer and the first layer crown high point group areas A1, A2, A3, . . . However, in order to clarify the distinction from each layer obtained in the second layer and below, which will be described later, on the original color image of FIG. Areas A1, A2, A3, . . . are each shown in red.

When the first layer crown high point group areas A1, A2, A3, . . . are extracted in this manner, Using a local maximum value filter (for example, 3 meshes × 3 meshes), extract the maximum value in each of the individual first layer tree crown high point group areas A1, A2, A3, ..., and extract the maximum value is the tree vertex of each tree crown (step S54). Since the vertex of the crown is the vertex of the tree having the crown, the tree height of the tree having the crown can be obtained by obtaining the vertex of the crown.

6 is a diagram showing tree vertices a1, a2, a3, . . . extracted from the first layer tree crown high point group areas A1, A2, A3, . The tree vertex obtained at the current processing target layer (in this case, the first layer) is indicated by a black dot.

In this way, when the process of extracting tree vertices a1, a2, a3, . Crown height)-D (layer width)×Ni (current processing target layer)=Hi (maximum crown height in the next processing target layer) (step S55).

The maximum tree crown height Hi in the next processing target layer is the maximum tree crown height (Hmax) in the survey target forest area (Hmax) = 25 m, the layer width (D) = 1 m, and the processing target layer Ni = 1, 25m−1m×1=24m, and the maximum tree crown height Hi in the next processing target layer is 24m. Then, it is determined whether or not Hi>0 (step S). In this case, since Hi=24 and Hi>0, Ni=Ni+1 is performed (step S57), and Ni=2 is set as the processing target. The hierarchy is set to the second hierarchy, and the process returns to the tree crown high point group data creation process for each hierarchy (step S53).

Here, the second layer is a layer with a tree crown height of 23 m to 24 m, and in this second layer, the tree crown high point group data creation process for each layer (step S53) is performed. That is, among the tree crown heights of each mesh, tree crown high point group data belonging to the second layer (referred to as second layer tree crown high point group data) is created. Then, the second layer crown high point group area corresponding to the second layer crown high point group data is displayed on the display.

FIG. 7 is a diagram showing the second layer crown high point group areas B1, B2, B3, . . . displayed on the display. In addition, in FIG. 7, in addition to the second layer crown high point group areas B1, B2, B3, . . . Tree vertices are also shown. In addition, in FIG. 7, the regions painted in gray are the second layer tree crown high point group regions B1, B2, B3, . . . indicates the presence of a tree whose vertex reaches the second level (23 m to 24 m). For example, in FIG. 7, looking at the state (cross-sectional view) of the group of trees contained in the black frame, among these groups of trees, the vicinity of the tree apex reaches the second layer indicated by a band. It can be seen that there are trees with

Since FIG. 7 is a monochrome image, in FIG. 7, the second layer and the second layer crown high point group areas B1, B2, B3, . . . However, in order to clarify the distinction from the tree crown high point group regions of each layer obtained in the other layers, on the original color image of FIG. The two-level tree crown high point group areas B1, B2, B3, . . . are shown in orange.

In FIG. 7, it is easy to distinguish between the newly extracted second layer crown high point group area and the already extracted first layer crown high point group areas A1, A2, A3, . Therefore, the already extracted first layer tree crown high point group areas A1, A2, A3, . . . are shown in white.

As shown in FIG. 7, among the second layer crown high point group areas B1, B2, B3, . , A2, A3, . It is displayed as a ring-shaped tree crown high point group area surrounding (the first layer tree crown high point group areas A1 and A2). Here, the crown high point cloud area extracted from the same crown as the already extracted crown high point cloud area is defined as the "extracted crown high point cloud area", and the new crown high point cloud area other than the "extracted crown high point cloud area" The tree crown high point group area extracted in 2 is referred to as a "newly extracted tree crown high point group area".

When the second layer crown high point group areas B1, B2, B3, . . . are extracted in this manner, Using a local maximum value filter (for example, 3 meshes × 3 meshes), the maximum value in each of the individual second layer crown high point group regions B1, B2, B3, ... is calculated, and the calculated maximum value is the tree vertex of each tree crown (step S54).

At this time, among the hierarchical tree crown high point cloud data corresponding to individual tree crowns created in the hierarchy (in this case, the second hierarchy) where the maximum value should be calculated, the hierarchical crown high point cloud data has already been created. Calculate the maximum value by excluding the tree crown high point group data by hierarchy created corresponding to the same tree crown as the existing tree crown. Taking FIG. 7 as an example, in the second layer crown high point group areas, for example, the crown high point group areas B4 and B5 are assumed to be already extracted crown high point group areas, and the already extracted crown height The point cloud areas B4, B5, etc. are not subject to tree vertex calculation.

The tree vertices b1, b2, b3, . . . extracted in the second layer crown high point group areas B1, B2, B3, . Since FIG. 7 is a monochrome image, in FIG. 7, tree vertices b1, b2, b3, . . . Each of the extracted tree vertices in the first layer tree crown high point group region is indicated by a black dot. The tree vertices in the tree crown high point group region of the layer (in this case, the first layer) are indicated by red dots.

When the process of extracting tree vertices b1, b2, b3, ... in the second layer crown high point group area is completed, Hmax (maximum tree crown height in the survey target forest area) - D (layer width) x Ni (current Processing target layer)=Hi (maximum crown height in the next processing target layer) (step S55) is performed. The maximum tree crown height Hi in the next processing target layer is the maximum tree crown height (Hmax) in the survey target forest area (Hmax) = 25 m, the layer width (D) = 1 m, and the processing target layer Ni = 2, 25m−1m×2=23m, and the maximum tree crown height Hi in the next processing target layer is 23m. Then, it is determined whether or not Hi>0 (step S). In this case, since Hi=23 and Hi>0, Ni=Ni+1=2+1 is performed (step S57), and as Ni=3, The process target layer is set to the second layer, and the process returns to the tree crown high point group data generation process for each layer S53.

Here, the third layer is a layer with a tree crown height of 22 m to 23 m, and in this third layer, the tree crown high point group data creation process for each layer (step S53) is performed. That is, among the tree crown heights of each mesh, tree crown high point group data belonging to the third layer (referred to as third layer tree crown high point group data) is created. Then, the third layer crown high point group area corresponding to the third layer crown high point group data is displayed on the display.

FIG. 8 is a diagram showing the third layer crown high point group areas C1, C2, C3, . . . displayed on the display. In addition, in FIG. 8, not only the third layer crown high point group areas C1, C2, C3, . . . Tree vertices are also shown. Further, in FIG. 8, the areas filled in with gray are the third layer tree crown high point group areas C1, C2, C3, . . . indicates the presence of a tree whose vertex reaches the third layer (22 m to 23 m). For example, in FIG. 8, looking at the state (cross-sectional view) of the group of trees contained in the black frame, among these groups of trees, the vicinity of the tree apex reaches the third layer shown in a belt shape. It can be seen that there are trees with

Since FIG. 8 is a monochrome image, in FIG. 8, the third layer and the third layer crown high point group areas C1, C2, C3, . . . However, in order to clarify the distinction from the tree crown high point group regions of each layer obtained in other layers, the third layer and the third layer shown in a band shape on the original color image of FIG. The three-level tree crown high point group areas C1, C2, C3, . . . are shown in yellow.

In FIG. 8, the newly extracted second layer crown high point group area, the already extracted first layer crown high point group areas A1, A2, A3, . . . Point cloud region B1. B2, B3, . B2, B3, . . . are shown in white.

When the third layer crown high point group areas C1, C2, C3, . . . are extracted in this manner, Using a local maximum value filter (for example, 3 meshes × 3 meshes), the maximum value in each of the individual third layer crown high point group regions C1, C2, C3, ... is extracted, and the extracted maximum value is the tree vertex of the tree (step S54).

In this case as well, among the layered tree crown high point cloud data corresponding to individual tree crowns created in the layer (in this case, the third layer) for which the maximum value should be calculated, the layered crown high point cloud data has already been created. The maximum value is calculated by excluding the hierarchical crown high point group data corresponding to the same crown as the crown of the tree. Taking FIG. 8 as an example, for example, the crown high point group areas C4 and C5 among the third layer tree crown high point group areas are assumed to be already extracted tree crown high point group areas. The high point group areas C4 and C5 are not subject to tree vertex calculation.

The tree vertices c1, c2, c3, . . In addition, since FIG. 8 is a monochrome image, in FIG. 8, each of the tree vertices extracted in the tree crown high point group regions of all layers up to this point is indicated by a black dot. In order to clarify the distinction between the tree vertices in the first layer crown high point group area and the tree vertices extracted in the second and first layer crown high point group areas, the original color image of FIG. , the tree vertices extracted in the second-layer and first-layer tree crown high point group regions are indicated by red circle points.

When the processing for extracting tree vertices c1, c2, c3, . The separate crown high point group data creation process (step S53) and the tree vertex calculation process (step S54) are successively repeated.Such processes may be performed until Hi=0, but not necessarily Hi=0. It is not necessary to do this until it reaches 0, and it can be set as appropriate according to the conditions of the survey target forest area. good.

FIG. 9 is a diagram showing the tree crown high point group areas for each layer from the first layer to the fifth layer and tree vertices in the tree crown high point group areas for each layer. In FIG. 9, the tree crown high point group areas for each layer (first to fifth layer tree crown high point group areas) extracted in each layer are shown in different colors for each layer. Since FIG. 9 is a monochrome image, it is difficult to read each layer by color, but in fact, on the color image that is the basis of FIG. The layers are indicated by color, such as orange, yellow for the third layer, yellowish green for the fourth layer, and green for the fifth layer. The maximum value (tree vertex) in the new tree crown high point group region extracted in each layer is indicated by a black dot. By sequentially performing such processing up to lower layers, a crown high point cloud image of each tree crown can be obtained for each layer.

FIG. 10 shows the tree crown high point group data creation process for each layer (step S53) and the tree vertex calculation process (step S54) sequentially from the first layer (top layer) to the bottom layer. FIG. 4 is a diagram showing a tree crown high point group area and tree vertices; By sequentially performing the tree crown high point group data creation process for each layer (step S53) and the tree vertex calculation process (step S54) from the first layer (top layer) to the bottom layer, the first layer (top layer ) to the lowest layer, and all crown apexes existing in the forest area to be surveyed can be extracted. In FIG. 10, the black circle points are tree vertices of individual tree crowns for each layer. Here, since the vertex of each crown is the vertex of each tree having that crown, the tree height of the tree having that crown can be obtained by obtaining the vertex of the crown. The tree height of each tree can be obtained from the noise-removed crown height image data from which noise has been removed by the averaging process (step S51).

In FIG. 10, the layered crown high point group regions extracted in each layer are shown in different colors for each layer. Since FIG. 10 is a monochrome image, it is difficult to read each layer by color. The layers are indicated by color, such as orange, yellow for the third layer, yellowish green for the fourth layer, and green for the fifth layer.

After all tree vertices existing in the survey target forest area are extracted in this way, the precise crown image data creation process (step S60) in the flowchart of FIG. 1 is performed. This precise canopy image data creation process is a process of creating precise canopy image data that enables highly accurate extraction of the canopy corresponding to each tree.

In the precise crown image data creation process, based on the crown vertices extracted for each layer, an area division algorithm is used to divide adjacent tree crowns from the crown height of the tree crown. Precise crown image data in which the crown is segmented is created. Specifically, using the tree apex of the tree crown extracted for each layer as a seed (Seed), the known region A detailed canopy image is created using the Watershed algorithm, which is a segmentation method. In the forest resource information calculation method according to the embodiment, noise-removed canopy height image data from which noise has been removed by the averaging process (step S51) shown in FIG. 4 is used as the z-value.

As mentioned above, the noise-removed canopy height image data is obtained by removing the extreme irregularities that exist in individual tree crowns (parts where branches and leaves are sparse compared to the surroundings, long branches and leaves, etc.). is smoothed. Therefore, when performing region segmentation by the Watershed algorithm, by using such noise-removed tree canopy height image data, region segmentation by the Watershed algorithm can be performed with high accuracy.

The Watershed Algorithm is an algorithm capable of classifying individual objects by extracting the boundaries of individual objects with high accuracy when a large number of closely spaced adjacent objects exist in a connected state. The Watershed algorithm is described in Publicly Known Document 2 shown below.

Public Document 2: Salman N. and Liu C. Q., “Image Segmentation and Edge Detection Based on Watershed Techniques,” International Journal of Computers and Applications, vol. 25, no. 4, pp. 258-263, 2003

The precise canopy image data creation process (step S60) creates precise canopy image data using the Watershed algorithm, and creates a precise canopy image corresponding to the precise canopy image data created by the precise canopy image data creation process (step S60). is displayed on the display.

FIG. 11 shows a detailed canopy image displayed on the display. As shown in FIG. 11, in the precise canopy image created using the Watershed algorithm, each canopy is segmented to surround the apex of each tree. In FIG. 11, areas surrounded by frame lines L are individual tree crowns.

In FIG. 11, the tree crown high point group regions for each layer extracted in each layer are shown in different colors for each layer. Since FIG. 11 is a monochrome image, it is difficult to read each layer by color. Layers are indicated by colors such as red, second layer orange, third layer yellow, fourth layer yellowish green, and fifth layer green.

As can be seen from the detailed canopy image shown in FIG. 11, the canopy corresponding to each tree obtained by the area division by the Watershed algorithm represents the canopy of the tree belonging to each hierarchy with high accuracy. It becomes a thing. As mentioned above, the Watershed algorithm is an algorithm that can classify each object by extracting the boundaries of each object with high precision when a large number of closely spaced objects exist in a connected state. is. Therefore, by using the Watershed algorithm, densely existing tree crowns can be classified as individual tree crowns with high accuracy regardless of the size of the tree crown (crown diameter and height).

When the precise crown image data is created in this manner, the precise crown information creation process (step S70) is performed based on the precise crown image data to obtain precise information (precise crown information) on the crown of each tree. ). The precise crown information created in the precise crown information creation process (step S70) includes at least the crown label number (serial number) corresponding to each tree, the crown position of the crown corresponding to each tree (crown center position), The crown diameter of the crown corresponding to each tree, the crown area of the crown corresponding to each tree, etc. can be exemplified, and in addition, the crown shape of the crown corresponding to each tree can be obtained as precise crown information. .

Subsequently, forest resource information calculation processing (step S80) is performed. The forest resource information calculation process calculates information on forest resources based on the precise crown information created in the precise crown information creation process (step S70). Examples of the information about forest resources calculated in the forest resource information calculation process (step S80) include the diameter at breast height of each tree and the volume of each tree. In addition to the above-mentioned diameter at breast height and timber volume of each tree, the tree height of each tree can be mentioned as the information about the forest resources. The tree height of each tree can be obtained based on the tree vertices calculated in the tree vertex calculation process (step S54).

In order to calculate information on these forest resources, the forest resource information calculation process (step S80) includes a breast height diameter calculation process (step S81) for calculating the breast height diameter of each tree, and a timber volume as a resource for each tree. A timber amount calculation process (step S82) and a forest resource information tabulation process (step S83) for tabulating the forest resource information are performed.

The diameter at breast height calculation process (step S81) calculates the diameter at breast height by a multiple regression equation from at least one of the crown area and crown diameter, which have a strong correlation with the diameter at breast height, and the tree height. Interpolation can also be used to estimate the DBH of unmeasured trees. From this, it is possible to calculate the breast-height diameter of all surveyed trees existing in the surveyed forest area. When obtaining the multiple regression equation, it is possible to obtain the variables of the multiple regression equation by selecting about 10 standard trees for each tree species in a field survey and measuring the crown area, crown diameter and tree height of the selected standard trees. preferable.

The timber volume determination process (step S82) obtains the timber volume from a two-variable timber volume formula of diameter at breast height and tree height. Note that a quick reference table is known in which the timber volume is obtained from the breast height diameter and the tree height, and the wood resources (timber volume) can be obtained from such a quick reference table. The quick reference table is described in the following documents (publicly known documents 3 and 4).

Publicly Known Document 3: Forestry Agency Planning Division, Tree Trunk Volume Table East Japan Edition, Japan Forestry Research Institute, page 334, 2003 Publicly Known Document 4: Forestry Agency Planning Division, Tree Trunk Volume Table West Japan Edition, Japan Forestry Research Institute, Page 320, 1970

The forest resource information aggregation process (step S83) aggregates the resource information of each tree such as the number of surveyed trees, crown diameter, crown area, breast height diameter, height and volume of each tree, that is, single tree resource information. . Then, the aggregated single tree resource information is registered in the attribute database. Single tree resource information registered in the attribute database can be displayed on the display as a list. The single tree resource information displayed as a list also includes XY coordinates representing the center position of the tree crown. Further, the number of survey target trees can be obtained based on the label numbers obtained by the precise crown information creation process (step S70).

By the way, the above-mentioned "tree crown diameter" represents the size of the horizontal spread of the tree crown when viewed from above, and the "breast height diameter" is the thickness of the trunk of the tree (in each tree, It represents the diameter of a human approximately at chest height). Since the crown and the trunk of a tree are not actually a perfect circle, it cannot be called a "diameter" in the strict sense. It is written as "diameter", and it is written as "breast height diameter" to represent the thickness of the trunk.

FIG. 12 is a diagram showing a list of single tree resource information displayed on the display. As shown in FIG. 12, as the single tree resource information of the forest trees to be surveyed, the crown number (label number) corresponding to each tree, the XY coordinates representing the center position of each tree crown, the crown diameter, the crown area, and the chest height. The diameter (denoted as "DBH" in FIG. 12), tree height, and volume are displayed in association with the crown number (label number).

In addition to the number of trees to be surveyed, the forest resource information aggregation process (step S83) includes the average crown diameter calculated for each tree, the maximum crown diameter calculated for each tree, and the minimum, mean crown area calculated for each tree, maximum and minimum crown area calculated for each tree, mean DBH diameter calculated for each tree, Mean value of DBH calculated for each tree, maximum and minimum DBH calculated for each tree, mean value of tree height calculated for each tree, for each tree Calculate the maximum and minimum values of the tree height calculated by the method, the average value of the timber volume calculated for each tree, and the maximum and minimum values of the timber volume calculated for each tree, Calculate the total value of timber volume, and calculate the amount of forest resources per hectare (ha) used in forest management (here, the number of trees and timber volume). It also performs processing to register in the attribute database as information indicating the outline of forest resources (forest resource outline information). Forest resource summary information registered in the attribute database can be displayed on the display as a list.

FIG. 13 is a diagram showing a list of forest resource summary information displayed on the display. As shown in Figure 13, the number of surveyed trees, the average crown diameter, the maximum and minimum crown diameters, the average crown area, the maximum and minimum crown areas, the average breast height diameter, and the breast height Average diameter, maximum and minimum diameter at chest height, average tree height, maximum and minimum tree height, average volume, maximum and minimum volume, and total volume are displayed. , and the amount of forest resources per hectare (ha) used in forest management (here, the number and volume of trees) are displayed. By displaying such forest resource summary information on the display, it is possible to grasp the summary of the forest resources in the surveyed forest area.

In this specification, the single tree resource information shown in FIG. 12 and the forest resource summary information shown in FIG. 13 are collectively referred to as forest resource information. By combining such forest resource information with existing forest survey database (for example, databases such as forest management ledgers maintained by prefectures), it is possible to grasp the current state of forests and plan forest management such as thinning. can be used for forest management. For example, when trying to thin a forest, it will be useful when simulating a thinning plan.

As described above, in the forest resource information calculation method according to the embodiment, after creating 3D point cloud data from aerial image data obtained by photographing an area including a survey target forest area using a drone or the like, the 3D point cloud data Create survey target forest area image data based on group data and geographical information, and based on the created survey target forest area image data, crown height image data representing tree crown heights existing in the survey target forest area to create Then, after creating noise-removed tree crown height image data by removing noise from the tree crown height image data by averaging processing, based on the noise-removed tree crown height image data, existing in each layer including the first layer The hierarchical tree crown high point cloud data corresponding to each tree crown is created for each layer, and the hierarchical crown high point cloud data corresponding to each tree crown created for each layer is used to create the tree crown corresponding to each tree crown. The maximum value of the high point cloud data is calculated.

Here, hierarchical crown high point cloud data is created for each layer, and from the hierarchical crown high point cloud data corresponding to each tree crown created for each layer, the hierarchical crown high point cloud corresponding to each tree crown is obtained. The process of calculating the maximum value of the data is, for example, by slicing each tree crown for each layer like a CT (Computed Tomography) scan to create tree crown high point cloud data for each layer. It is a process of calculating the maximum value from the tree crown high point group data for each hierarchy obtained. By performing such processing, the maximum value of individual tree crowns existing in each layer can be calculated for each layer. Note that the “tree top of tree crown” calculated for each layer is the tree top of the tree having the tree crown, and the tree height of the tree can be obtained from the calculated tree top.

In this way, after calculating the maximum value (tree vertex) of individual tree crowns existing in each layer, each tree crown A watershed algorithm is used to create precise crown image data in which adjacent crowns are segmented into individual crowns from the crown heights of .

The precise canopy image data created in this manner is precise canopy image data that enables highly accurate extraction of individual canopies corresponding to the trees to be surveyed. For this reason, the precise crown information created based on the precise crown image data becomes highly accurate crown information corresponding to each tree of the survey target trees, and information on forest resources (forest resource information ), it is possible to obtain highly accurate forest resource information. As a result, it is possible to provide high-precision forest resource information with little measurement error and high objectivity for the entire forest area to be surveyed, for each subdivision, and for any range, so it can be used for forest management such as planning forest operations such as thinning. .

In addition, in the embodiment, since a general camera and a popular drone can be used as shooting equipment for obtaining aerial image data, a system for calculating forest resources is inexpensive and easy. can be realized.

In addition, since aerial image data obtained by photographing the surveyed forest area with a camera is used, understory vegetation (shrubs and grasses not included in the survey) existing under the tree canopy (between the tree canopy and the ground) etc.) is less likely to be captured as image data, the effect of reducing noise processing for removing understory vegetation as noise can also be obtained.

The forest resource information calculation method 1 according to the embodiment has been described above. Subsequently, the forest resource information calculation device according to the embodiment will be described.

FIG. 14 is a diagram for explaining the forest resource information calculation device 1 according to the embodiment. The forest resource information calculation device 1 according to the embodiment is a device for performing the processing of each step shown in FIGS. , a three-dimensional point cloud data creation unit 20 having the function of performing the processing of step S20 in FIG. 1, a survey target forest area image data creation unit 30 having the function of performing the processing of step S30 in FIG. A tree crown height image data generation unit 40 having a function of performing the processing of step S40, a hierarchical tree vertex extraction unit 50 having a function of performing the processing of step S50 in FIG. 1, and a function of performing the processing of step S60 in FIG. an accurate crown image data creation unit 60 having a precise crown image data creation unit 70 having the function of performing the processing of step S70 in FIG. 1; a forest resource information calculation unit 80 having the function of performing the processing of step S80 in FIG. have

The hierarchy-specific tree vertex extraction unit 50 also includes components for performing a plurality of processes, which will be described later with reference to FIG. 15 . In addition, the forest resource information calculation unit 80 includes a chest height diameter calculation unit 81 having a function of performing the processing of step S81 in FIG. and a forest resource information tallying unit 83 having a function of performing the processing of step S83 in .

FIG. 15 is a diagram showing the configuration of the hierarchical tree vertex extraction unit 50 shown in FIG. As shown in FIG. 15, the hierarchical tree vertex extracting unit 50 includes an averaging processing unit 51 for performing the averaging processing in step S51 in FIG. 4 and an initial value setting processing in step S52 in FIG. The initial value setting unit 52, the hierarchical crown high point group data creation unit 53 for performing the hierarchical crown high point group data creation processing in step S53 in FIG. 4, and the tree vertex calculation processing in step S54 in FIG. and a calculation processing unit 58 for performing calculations such as Hmax−D×Ni=Hi in step S55, Hi>0 in step S56, and Ni=Ni+1 in step S57 in FIG. there is

As described above, each component of the forest resource information calculation device 1 according to the embodiment shown in FIGS. ), the forest resource information calculation device 1 according to the embodiment can obtain the same effects as those obtained by the forest resource information calculation method according to the embodiment described above. Note that the description of the processing contents of the forest resource information calculation device 1 according to the embodiment will be omitted.

Further, the forest resource information calculation device 1 according to the embodiment has the functions of each of the components (see FIGS. 14 and 5) included in the forest resource information calculation device 1 installed as a computer program. By giving predetermined data to each component, the function of each component is executed by computer software.

Such software includes a survey target forest area image data creation processing step of creating survey target forest area image data based on three-dimensional point cloud data corresponding to an area including the survey target forest area and geographical information, A crown height image data creation processing step for creating crown height image data representing the crown height of each tree existing in the survey target forest area based on the survey target forest area image data, and crown height image data. Among the tree crown heights obtained from the above, the layer having the maximum tree crown height is defined as the first layer, and a plurality of layers having a layer width of a predetermined length starting from the maximum tree crown height are set from the first layer toward the ground surface. and a hierarchical tree vertex extraction processing step of extracting tree vertices of a tree crown having tree vertices in each layer including the first layer when the tree vertices are present in each layer.

It should be noted that the present invention is not limited to the above-described embodiments, and various modifications can be made without departing from the gist of the present invention. For example, modified implementations as shown below are also possible.

(1) In the above embodiment, the trees to be surveyed are planted trees such as larch, Japanese red pine, Japanese cedar, and cypress. However, the trees to be surveyed are not limited to the above planted trees. Instead, for example, when there is a forest in which beech or oak grow in clusters within a predetermined area, the beech or oak can also be used as the trees to be surveyed.

(2) In the above embodiment, a drone is used to obtain aerial image data, but the invention is not limited to drones. For example, an aircraft may be used to obtain aerial image data. .

(3) In the above embodiment, SFM (Structure from Motion), which is a three-dimensional shape restoration technique, is used to create a distortion-free map-corrected orthorectified image from aerial image data. Although point cloud data is created, three-dimensional point cloud data may be obtained based on laser measurement data obtained by irradiating laser light from above.

Reference Signs List 1 Forest resource information calculation device 10 Aerial data input unit 20 Three-dimensional point cloud creation unit 30 Investigation target forest area image data creation unit 40 Canopy height Image data creating unit 50 Hierarchical tree vertex extracting unit 51 Averaging processing unit 52 Initial value setting unit 53 Hierarchical crown high point group data creating unit 54 Tree vertex calculation unit 58 Operation processing unit 60 Precise crown image data creation unit 70 Precise crown information creation unit 80 Forest resource information calculation unit 81 Chest height diameter Calculation unit 82... Timber calculation unit 83... Forest resource information aggregation unit S10... Aerial photograph data input process S20... Three-dimensional point group creation process S30... Survey target forest area Image data generation processing S40 Tree crown height image data generation processing unit S50 Hierarchical tree vertex extraction processing S51 Averaging processing S52 Initial value setting processing S53 Hierarchical crown High point group data creation process, S54... tree vertex calculation process, S55... calculation process of Hmax-DxNi=Hi, S56... calculation process of Hi>0, S57... Ni=Ni+1 Arithmetic processing, S60... Precise crown image data creation process, S70... Precise tree crown information creation process, S80... Forest resource information calculation process, S81... Breast height diameter calculation process, S82... Wood quantity calculation process , S83 . , B2, B3, . . . 2nd layer crown high point group area, b1, b2, b3 . High point group area, c1, c2, c3 . hierarchy to be processed, Hi ... the maximum crown height in the next hierarchy to be processed

Claims (9)

A forest resource information calculation device for creating forest resource information about a survey target forest area,
a survey target forest area image data creation unit that creates survey target forest area image data based on three-dimensional point cloud data corresponding to an area including the survey target forest area and geographical information;
a tree crown height image data creation unit that creates tree crown height image data representing the crown height of each tree existing in the survey target forest area based on the survey target forest area image data;
Among the tree crown heights obtained from the tree crown height image data, the layer having the maximum tree crown height is defined as the first layer, and the layer having a layer width of a predetermined length starting from the maximum tree crown height is selected from the first layer to the ground surface. a hierarchical tree vertex extraction unit that extracts tree vertices of a tree crown having tree vertices in each hierarchy including the first hierarchy when a plurality of tree vertices are set toward each hierarchy,
The crown height image data creation unit creates meshed digital surface model data and digital elevation model data based on the survey target forest area image data, and generates meshed digital surface model data and digital elevation model data for each mesh. Create crown height image data in which the crown height is obtained for each mesh by taking the difference from the model data,
In the hierarchical tree vertex extraction unit,
Using the tree crown height image data obtained by obtaining the tree crown height for each mesh, layered tree crown high point group data corresponding to individual tree crowns existing in each layer including the first layer is created for each layer. a tree crown high point cloud data creation unit for each hierarchy;
From the hierarchical crown high point cloud data corresponding to the individual tree crowns created by the hierarchical crown high point cloud data creation unit, the maximum value of the hierarchical crown high point cloud data corresponding to the individual tree crowns is a tree vertex calculation unit that calculates the tree vertex of the tree crown corresponding to the tree crown high point group data by hierarchy, and sets the calculated maximum value to the tree vertex of the tree crown corresponding to the tree crown high point group data by hierarchy;
A forest resource information calculation device characterized by including In the forest resource information calculation device according to claim 1 ,
When calculating the maximum value, the tree vertex calculation unit selects a tree crown for which the hierarchical crown high point group data has already been created among the hierarchical crown high point group data corresponding to the individual tree crowns. and calculating the maximum value while excluding hierarchical tree crown high point group data corresponding to the same tree crown. In the forest resource information calculation device according to claim 1 or 2 ,
In the hierarchical tree vertex extraction unit,
Each mesh of the tree crown height image data is set as a mesh to be processed, an averaging filter is superimposed on the mesh to be processed, and the average value of the tree crown heights of each mesh included in the averaging filter is the crown height of the mesh to be processed. It further includes an averaging processing unit that creates noise-removed tree canopy height image data by sequentially performing the processing for each processing target mesh,
The tree crown height point cloud data creation unit for each layer uses the noise-removed tree crown height image data created by the averaging processing unit as the tree crown height image data obtained by obtaining the tree crown height for each mesh, and uses the tree crown height image data to obtain the tree crown height for each mesh. A forest resource information calculation device characterized by creating separate crown high point group data. In the forest resource information calculation device according to any one of claims 1 to 3 ,
The forest resource information calculation apparatus, wherein the layer width is set to a length in the range of 0.2 m to 2.0 m. In the forest resource information calculation device according to any one of claims 1 to 4 ,
The three-dimensional point cloud data is created by SFM (Structure from Motion), which is a three-dimensional shape restoration technology, from photographed image data obtained by overlapping and shifting photographing locations from the sky. A forest resource information calculation device characterized by: In the forest resource information calculation device according to any one of claims 1 to 4 ,
A forest resource information calculation apparatus, wherein the three-dimensional point cloud data is created based on laser measurement data obtained by irradiating a laser beam from the sky. In the forest resource information calculation device according to any one of claims 1 to 6 ,
Based on the tree vertices of the tree crown extracted for each of the above-mentioned layers, a precise tree crown in which individual tree crowns are segmented using a region division algorithm that divides adjacent tree crowns from the crown height of the tree crown. A forest resource information calculation apparatus, further comprising a precise canopy image data creation unit for creating image data. In the forest resource information calculation device according to claim 7 ,
The forest resource information calculation apparatus, wherein the area division algorithm is a Watershed algorithm. In the forest resource information calculation device according to claim 7 or 8 ,
a precise crown information creation unit that creates precise crown information corresponding to each tree from the precise crown image data created by the precise crown image data creation unit;
a forest resource information calculation unit that creates forest resource information about the survey target forest area based on the precise crown information corresponding to each tree created by the precise crown information creation unit;
further having
The precise crown information corresponding to each tree includes at least a crown label number corresponding to each tree, a crown position of the crown corresponding to each tree, a crown diameter of the crown corresponding to each tree, and each tree. contains the crown area of the crown corresponding to
The Forest Resource Information Calculation Department has
a chest-height diameter calculator that calculates the diameter of each tree at approximately the height of a person's chest as the chest-height diameter of each tree;
a timber calculation unit for calculating the timber volume of each tree;
The height of each tree obtained based on the diameter at breast height of each tree, the volume of each tree, the precise crown information created by the precise crown information creating unit, and the tree vertices extracted by the hierarchical tree apex extracting unit A forest resource information calculation device characterized by comprising a forest resource information totaling unit that performs processing for totalizing the forest resource information.

Images