Tree template generation system

Abstract

The object of the present invention is to solve the problems of the conventional methods, namely, to provide a tree template generation system that can generate a tree template in which missing regions of branches and leaves are filled in, using measured point cloud data. [Solution] The tree template generation system of the present invention is a system that generates a tree template that can be placed in three-dimensional space using a "tree point cloud", and comprises a trunk model generation means, a branch and leaf model generation means, a missing area setting means, a missing area completion means, and a tree template storage means. Of these, the missing area setting means is a means for setting branch and leaf missing areas where branches and leaves are missing during measurement. The missing area completion means is a means for completing branch and leaf missing areas with a part of the branch and leaf model.

Description

Technical Field

[0001] The present invention relates to a technology for generating a template that can be arranged in a viewer representing, for example, a three-dimensional space. More specifically, the present invention relates to a tree template generation system that generates a tree template based on actually measured three-dimensional measurement points.

Background Art

[0002] When the maintenance of rivers, roads, etc. is planned, it may be necessary to explain the landscape after maintenance to the neighboring residents and seek mutual agreement. In that case, it is necessary for the residents to understand the situation after the maintenance of the river, etc. Conventionally, an explanation has been made using a colored design drawing such as a plan view, a cross-sectional view, or a longitudinal-sectional view. However, a certain degree of knowledge and familiarity are required to understand the design drawing, and the residents who receive the explanation may not necessarily understand it. Further, even if the design drawing can be understood, it is difficult to imagine the situation after maintenance as an actual thing, and it has not been easy for the residents to examine and judge the validity of the maintenance plan at the explanation meeting.

[0003] Therefore, in order to convey the landscape after maintenance more specifically, various measures have been taken, such as creating a photomontage obtained by subjecting a photograph of the current situation to composite processing, commissioning a professional painter to draw a completed drawing, creating computer graphics based on a design drawing, creating a miniature model, etc.

[0004] On the other hand, in recent years, as topographic measurement technology and information and communication technology (ICT) have advanced, it has become possible to process large amounts of measurement data. For example, with MMS (Mobile Mapping System), a huge number of 3D measurement points (so-called point cloud data) can be acquired in a single measurement, and by processing this point cloud data, a topographic model can be generated. Consequently, viewers that place and draw topographic models in 3D space have also come into use. Such viewers offer superior visual clarity compared to design drawings, photomontages, and computer graphics, and are therefore increasingly being used in situations where river and other infrastructure improvements are explained to residents.

[0005] Incidentally, when improving the river environment, plans to plant trees on riverbanks and other areas are often included. In this case, using a viewer or similar tool for explanation allows for the virtual placement and display of trees, and the size, species, and placement of the trees can be changed on the spot, which is extremely useful in reaching a consensus between residents and the planners. For this purpose, it is desirable to prepare virtual data (i.e., templates) of various trees in advance. Therefore, various techniques for creating tree templates have been proposed to date. For example, Patent Document 1 proposes a technique for representing trees by randomly generating points around a set sphere. [Prior art documents] [Patent Documents]

[0006] [Patent Document 1] Japanese Patent Publication No. 2014-85836 [Overview of the Initiative] [Problems that the invention aims to solve]

[0007] As a viewer used in various informational sessions, there are high expectations for game engines, which have seen particularly remarkable progress recently. One reason for this is the belief that various geographical features, including trees, can be represented more realistically by using a game engine. However, since trees and other features rendered by a game engine are not created based on actual measurements, but rather are essentially drawn data, there are limitations to how realistically they can be represented.

[0008] Therefore, it is conceivable to use point cloud data acquired by MMS or similar systems to draw trees. Specifically, a tree model can be created by using the point cloud data as is, and this tree model can be edited and stored as a template. For example, the tree model shown in Figure 7 is drawn directly from point cloud data acquired by MMS, and it is represented with extreme realism.

[0009] However, measurement methods like MMS have the characteristic that while they can acquire a dense number of measurement points on the part facing the measuring instrument, they have difficulty acquiring sufficient measurement points on the back side. Therefore, although the tree model shown in Figure 7 appears at first glance to accurately represent an actual tree, in reality, as shown in Figure 8(a), when viewed from above, it can be seen that the branches and leaves on the back side are missing (within the dashed lines shown in the figure). Alternatively, as shown in Figure 8(b), the tree model may be created with only the upper part of the branches and leaves that should actually be present missing (within the dashed lines shown in the figure).

[0010] The object of the present invention is to solve the problems of the conventional methods, namely, to provide a tree template generation system that can generate a tree template in which missing regions of branches and leaves are filled in, using measured point cloud data. [Means for solving the problem]

[0011] The present invention focuses on generating a tree template using measured point cloud data and supplementing missing areas of branches and leaves with that point cloud data, and is based on an idea that has not been seen before.

[0012] The tree template generation system of the present invention is a system that generates a tree template that can be placed in three-dimensional space using a "tree point cloud (a collection of three-dimensional measurement points obtained by measuring a tree)", and comprises a trunk model generation means, a branch and leaf model generation means, a missing region setting means, a missing region completion means, and a tree template storage means. Of these, the trunk model generation means is a means for generating a trunk model based on a "trunk point cloud (a point cloud from the tree point cloud that corresponds to the trunk of a tree)", and the branch and leaf model generation means is a means for generating a branch and leaf model based on a "branch and leaf point cloud (a point cloud from the tree point cloud that corresponds to the branches and leaves of a tree)". Furthermore, the missing region setting means is a means for setting a branch and leaf missing region where branches and leaves were missing during measurement, the missing region completion means is a means for completing the branch and leaf missing region with a part of the branch and leaf model, and the tree template storage means is a means for storing a tree template consisting of a trunk model and a branch and leaf model in which the branch and leaf missing region has been completed.

[0013] The tree template generation system of the present invention may further include a trunk diameter calculation means and a tree template acquisition means. The trunk diameter calculation means is a means for calculating the trunk diameter of a trunk model, and the tree template acquisition means is a means for reading other tree templates having a trunk diameter equivalent to the trunk diameter from a tree template storage means. In this case, the missing region completion means completes the missing branch and leaf regions using a part of the branch and leaf model that constitutes the other tree template acquired by the tree template acquisition means.

[0014] The tree template generation system of the present invention can also complement missing branch and leaf regions using a portion of the branch and leaf model that constitutes the same tree template. In this case, the means for complementing missing regions moves a portion of the branch and leaf model so as to be symmetrical with respect to the trunk model, or moves it so as to be point-symmetrical with respect to a reference point set in the trunk model, and then complements the missing branch and leaf regions.

[0015] The tree template generation system of the present invention can also store branch and leaf models generated from tree point clouds obtained by measurements from above. In this case, the trunk model generation means generates a trunk model using tree point clouds obtained by measurements from the ground, and the branch and leaf model generation means generates a branch and leaf model using tree point clouds obtained by measurements from the ground. The missing area completion means reads the branch and leaf models related to the aerial measurements from the tree template storage means and completes the missing branch and leaf areas with a part of the read-out branch and leaf model.

[0016] The tree template generation system of the present invention can also enlarge or reduce a portion of the branch and leaf model and then fill in the missing branch and leaf areas. [Effects of the Invention]

[0017] The tree template generation system of the present invention has the following effects: (1) It is possible to place more realistic trees in a three-dimensional space, which allows for a more concrete visualization of, for example, river improvement plans. (2) Because it uses only measured point cloud data and does not require complex editing processes, tree templates can be generated at low cost. (3) Even if areas with missing branches and leaves are found, there is no need to perform additional measurements, and in this respect as well, tree templates can be generated at a low cost. [Brief explanation of the drawing]

[0018] [Figure 1]A step diagram schematically showing the general procedure until a tree template is generated using the tree template generation system of the present invention. [Figure 2] A block diagram showing the main configuration of the tree template generation system of the present invention. [Figure 3] (a) is a model diagram schematically showing the "branch and leaf cross-section" set in the tree model, and (b) is a cross-sectional view showing the branch and leaf defect area included in the branch and leaf cross-section. [Figure 4] (a) is a model diagram schematically showing the situation of selecting a complementary branch and leaf model from the branch and leaf models of the same tree model, and (b) is a model diagram schematically showing the situation of complementing the branch and leaf defect area with the selected complementary branch and leaf model. [Figure 5] A model diagram schematically showing the situation where the defect area is complemented by the complementary model branch and leaf model selected from other tree templates. [Figure 6] A model diagram schematically showing the situation where a branch and leaf model based on aerial measurement is selected to complement the branch and leaf defect area of the branch and leaf model based on ground measurement. [Figure 7] An image diagram showing a tree model drawn with point cloud data acquired by MMS. [Figure 8] (a) is an image diagram of a tree model with missing branches and leaves on the back side viewed from above, and (b) is an image diagram of a tree model with missing upper branches and leaves viewed from the side.

Embodiments for Carrying Out the Invention

[0019] An example of an embodiment of the tree template generation system of the present invention will be described based on the drawings.

[0020] The tree template generation system of the present invention generates a template of a tree (hereinafter simply referred to as "tree template") that can be arranged in a viewer or the like that represents a three-dimensional space, and is an invention that generates a tree template based on a large number of three-dimensional measurement points (hereinafter referred to as "point cloud data") actually measured.

[0021] Figure 1 is a schematic step diagram illustrating the general procedure for generating a tree template using the tree template generation system of the present invention. In generating a tree template using the tree template generation system, point cloud data corresponding to trees (hereinafter specifically referred to as "tree point cloud") is extracted from the acquired point cloud data, and a three-dimensional model of a tree (hereinafter simply referred to as "tree model") is generated from this tree point cloud.

[0022] Next, a three-dimensional model of the tree trunk (hereinafter simply referred to as the "tree trunk model") is generated based on the point cloud data corresponding to the tree trunk (hereinafter referred to as the "tree trunk point cloud"), and a three-dimensional model of the branches and leaves (hereinafter simply referred to as the "branch and leaf model") is generated based on the point cloud data corresponding to the branches and leaves (hereinafter referred to as the "branch and leaf point cloud") within the tree point cloud. Various conventional techniques can be used to extract the tree point cloud from the acquired point cloud data and to generate the tree trunk model and branch and leaf model. For example, the paper "ROADSIDE TREE EXTRACTION AND DIAMETER ESTIMATION WITH MMS LIDAR BY USING POINT-CLOUD IMAGE (G. Takahashi, H. Masuda)" proposes a technique for classifying trees using point cloud features, creating high-precision wireframes using point cloud images, and then extracting columnar structures (i.e., tree trunks) using these wireframes. Alternatively, the website "https: / / www.docswell.com / s / HoloLab / Z4Q886-holoconf24_b3#p14" proposes a technique for separating leaves, branches, and trunks by processing point cloud reflectance with a classifier.

[0023] When generating the trunk model and branch / leaf model, regions where branches and leaves that should actually exist were not measured (hereinafter referred to as "branch / leaf missing regions") are defined in the branch / leaf model. Then, a partial branch / leaf model that can complement the branch / leaf missing regions (hereinafter referred to as "complementary branch / leaf model") is selected, and the branch / leaf missing regions of the branch / leaf model are complemented using this complementary branch / leaf model. When selecting the complementary branch / leaf model, as described later, it is possible to select from branch / leaf models that make up the same tree model, or from branch / leaf models that make up other tree models. Finally, a "tree template" is generated using the branch / leaf model complemented by the complementary branch / leaf model and the trunk model.

[0024] Figure 2 is a block diagram showing the main components of the tree template generation system 100 of the present invention. As shown in this figure, the tree template generation system 100 of the present invention is composed of a trunk model generation means 101, a branch and leaf model generation means 102, a missing area setting means 103, a missing area completion means 104, and a tree template storage means 109. It can also be further composed of a trunk diameter calculation means 105, a tree template acquisition means 106, a drawing means 107, a point cloud data storage means 108, and the like.

[0025] Of the various means constituting the tree template generation system 100, the trunk model generation means 101, the branch and leaf model generation means 102, the missing area setting means 103, the missing area completion means 104, the trunk diameter calculation means 105, the tree template acquisition means 106, and the drawing means 107 can be manufactured as dedicated components or a general-purpose computer device can be used. In other words, the computer device is made to perform calculations according to a predetermined program, thereby performing processing specific to each means. This computer device can consist of a personal computer (PC), a tablet PC such as the iPad (registered trademark), a mobile terminal including a smartphone, or a PDA (Personal Data Assistance). The computer device is equipped with a processor such as a CPU, memory such as ROM or RAM, and may also include input means such as a mouse or keyboard and a display.

[0026] Furthermore, the point cloud data storage means 108 and the tree template storage means 109 can utilize the storage device of a general-purpose computer (e.g., a personal computer) or be built on a database server. When built on a database server, it can be located on a local network (LAN: Local Area Network) or on a cloud server that stores data via the internet.

[0027] The following describes in detail each of the main elements that constitute the tree template generation system 100 of the present invention.

[0028] (Point cloud data storage means) The point cloud data storage means 108, which constitutes the tree template generation system 100, stores measured point cloud data. The point cloud data storage means 108 can store point cloud data obtained by measurements such as MMS or ground-mounted laser measurement (hereinafter simply referred to as "ground measurement"), or by measurements such as laser measurement and photogrammetry using aircraft or so-called drones (UAVs: Unmanned Aerial Vehicles) (hereinafter simply referred to as "aerial measurement"). Furthermore, it can store point cloud data derived from various measurements, such as storing both point cloud data from ground measurement and point cloud data from aerial measurement.

[0029] (Tree template storage means) The tree template storage means 109, which constitutes the tree template generation system 100, is a means for storing tree templates as well as tree models, trunk models, and branch and leaf models. It is preferable to store the tree models, trunk models, and branch and leaf models in association with the respective tree templates. Furthermore, it is preferable to store the point cloud data measurement methods (such as ground measurements and aerial measurements) used to generate the tree models, trunk models, and branch and leaf models in association with the respective tree models, trunk models, and branch and leaf models.

[0030] (Tree trunk model generation means and branch / leaf model generation means) The tree template generation system 100 includes a trunk model generation means 101 which generates a trunk model based on the trunk point cloud of the tree point cloud, and a branch and leaf model generation means 102 which generates a branch and leaf model based on the branch and leaf point cloud of the tree point cloud. In generating the trunk model with the trunk model generation means 101 and generating the branch and leaf model with the branch and leaf model generation means 102, various conventional techniques can be used as described above.

[0031] (Means for setting missing regions) The missing area setting means 103, which constitutes the tree template generation system 100, is a means for setting missing areas in the branch and leaf model. When setting missing areas in the branch and leaf using the missing area setting means 103, it is possible to specify that the setting is done by operator operation. Specifically, the tree model (trunk model and branch and leaf model) is displayed in a 3D viewer or 3D CAD (Computer Aided Design) that can represent three-dimensional space, and the operator sets the desired missing areas in the branch and leaf model by operating a pointing device (mouse, touch panel, pen tablet, touchpad, trackpad, trackball, etc.) or keyboard while visually observing the branch and leaf model. However, this missing area in the branch and leaf is set as a three-dimensional area.

[0032] The missing area setting means 103 can be configured to automatically set the missing branch and leaf areas, instead of (or in addition to) the configuration to set the missing branch and leaf areas by operator operation. The procedure for the missing area setting means 103 to automatically set the missing branch and leaf areas will be described below with reference to Figure 3. First, as shown in Figure 3(a), a central axis (hereinafter referred to as the "trunk axis") is set for the tree trunk model, and multiple cross-sections (hereinafter referred to as "branch and leaf cross-sections") are set at predetermined intervals along the trunk axis. These branch and leaf cross-sections are planes perpendicular to the trunk axis, and when viewed from above, as shown in Figure 3(b), they are circles centered on the trunk axis. The radius of the branch and leaf cross-section should be set according to the position of the branches and leaves, such as based on the branch and leaf furthest from the trunk axis.

[0033] When multiple branch and leaf cross-sections are set along the trunk axis, the area occupied by branches and leaves within the branch and leaf cross-section (hereinafter referred to as the "branch and leaf area") is calculated, as shown in Figure 3(b), and the area excluding the branch and leaf area (hereinafter referred to as the "missing area") is also calculated. At this time, the branch and leaf area and the missing area can be calculated using a branch and leaf point cloud, or they can be calculated using various conventional image recognition techniques. When the missing area occupied by the branch and leaf cross-section exceeds a predetermined threshold, it is determined that the missing area constitutes a branch and leaf missing region, and the branch and leaf missing region is calculated by summing (integrating) the missing areas related to the branch and leaf missing region in the direction of the trunk axis.

[0034] (Means for filling in missing areas) The missing region completion means 104 that constitutes the tree template generation system 100 is a means of selecting a completion branch and leaf model and then completing the missing branch and leaf regions using that completion branch and leaf model. As previously described, there are two main methods for selecting a completion branch and leaf model: one that selects from branch and leaf models that constitute the same tree model (hereinafter referred to as the "same model type") and another that selects from branch and leaf models that constitute other tree models (hereinafter referred to as the "other model type"). The same model type and the other model type will be explained in order below.

[0035] Figure 4 schematically shows how the missing region completion means 104 completes the missing branch and leaf regions using the same model type. (a) is a model diagram showing the process of selecting a completion branch and leaf model from the branch and leaf models of the same tree model, and (b) is a model diagram showing the process of completing the missing branch and leaf regions using the selected completion branch and leaf model.

[0036] In the same model type, as shown in Figure 4(a), the missing area completion means 104 first selects a complementary branch and leaf model that can complete the missing branch and leaf area from the branch and leaf models that constitute the same (its own) tree model. When setting the missing branch and leaf area using the missing area completion means 104, it is possible to specify that the setting is done by operator operation. Specifically, the tree model is displayed in a 3D viewer or 3D CAD, and the operator selects the desired complementary branch and leaf model by visually inspecting the branch and leaf models and operating a pointing device or keyboard, etc. However, this complementary branch and leaf model is set as a 3D region.

[0037] The missing region completion means 104 can also be configured to automatically select a completion branch and leaf model, instead of (or in addition to) the specification of setting branch and leaf missing regions by operator operation. For example, a 3D branch and leaf missing region can be used as a template, and a 3D shape that matches that branch and leaf missing region (template) can be detected from the branch and leaf model, and the detected 3D shape can be selected as the completion branch and leaf model.

[0038] When a complementary branch and leaf model is selected from the same tree model, the missing area completion means 104 completes the missing branch and leaf area using the complementary branch and leaf model, as shown in Figure 4(b). At this time, the orientation of the missing branch and leaf area and the complementary branch and leaf model are matched before completion. For example, a part of the branch and leaf model can be duplicated as the complementary branch and leaf model, and then moved to be symmetric with respect to the trunk axis before completing the missing branch and leaf area. Alternatively, the complementary branch and leaf model can be rotated and moved by a desired angle around the trunk axis before completing the missing branch and leaf area, or it can be rotated and moved to be point-symmetric with respect to a reference point set on the trunk axis before completing the missing branch and leaf area.

[0039] Next, with reference to Figure 5, the procedure by which the missing area completion means 104 completes the missing branch and leaf areas using other model types will be explained. In other model types, first the trunk diameter calculation means 105 calculates the diameter of the trunk model that constitutes the tree model (hereinafter referred to as "trunk diameter"). Then the tree template acquisition means 106 reads out other tree templates from the tree template storage means 109 based on that trunk diameter. For example, the tree template acquisition means 106 can read out a tree template that has a trunk diameter equivalent to the trunk diameter of the tree model in which the missing branch and leaf areas are to be completed. Here, "equivalent" means that the trunk diameters of both are the same or approximate, and they can be judged as "equivalent" when the difference (e.g., absolute value) between the trunk diameter of one and the trunk diameter of the other falls below a predetermined threshold.

[0040] When another tree template is read from the tree template storage means 109, the missing area completion means 104 selects a complementary branch and leaf model from the branch and leaf models that constitute the other tree template that can complete the missing branch and leaf region. When setting the missing branch and leaf region using the missing area completion means 104, it can be set by operator operation, similar to the same model type, or it can be set to automatically select a complementary branch and leaf model. Once a complementary branch and leaf model is selected, the missing area completion means 104 completes the missing branch and leaf region using that complementary branch and leaf model. At this time, similar to the same model type, the missing branch and leaf region can be completed by moving the complementary branch and leaf model so as to be symmetrical with respect to the trunk axis, rotating the complementary branch and leaf model by a desired angle around the trunk axis, or rotating it so as to be point-symmetrical around a reference point set on the trunk axis.

[0041] As previously described, the tree template storage means 109 can store branch and leaf models, and moreover, it can store them in association with the point cloud data measurement method used to generate those branch and leaf models. Therefore, in other model types, it is also possible to select complementary branch and leaf models from branch and leaf models derived from other measurement methods. For example, in Figure 6, in order to complement the branch and leaf missing regions of the branch and leaf model based on ground measurements, a branch and leaf model based on aerial measurements is selected as a complementary branch and leaf model. Specifically, the tree template acquisition means 106 selects and reads out a branch and leaf model based on aerial measurements from the tree template storage means 109, and the missing region complementation means 104 uses that branch and leaf model as a complementary branch and leaf model to complement the branch and leaf missing regions.

[0042] The missing region completion means 104 can also be configured to select a completion branch / leaf model that is similar to the missing branch / leaf region, regardless of whether it is the same model type or a different model type. In this case, the missing region completion means 104 calculates the similarity ratio between the selected completion branch / leaf model and the missing branch / leaf region, and then enlarges or reduces the completion branch / leaf model according to that similarity ratio before completing the missing branch / leaf region.

[0043] (Drawing means 107) The drawing means 107 that constitutes the tree template generation system 100 represents a three-dimensional space by using, for example, a 3D viewer or 3D CAD, and is a means that can place a tree template at a desired position and orientation within that three-dimensional space. The drawing means 107 can also display the tree template in color. In this case, it is preferable to generate the tree template by first creating a colored tree point cloud based on, for example, the reflectance intensity obtained from laser measurement or a separately acquired color photograph. [Industrial applicability]

[0044] The tree template generation system of the present invention can be used when explaining various plans such as river maintenance, road improvements, and the construction of new parks. Considering that the present invention enables the realization of river maintenance and other infrastructure projects only after agreement has been reached with nearby residents, it can be said that the present invention is not only industrially applicable but also has the potential to make a significant contribution to society. [Explanation of symbols]

[0045] 100 Tree template generation system of the present invention 101 (Tree template generation system) Tree trunk model generation means 102 (Tree template generation system) Branch and leaf model generation means 103 (Means for setting missing regions in the tree template generation system) 104 (Means for filling in missing areas of the tree template generation system) 105 (Tree template generation system) Trunk diameter calculation means 106 (Means for obtaining tree templates in the tree template generation system) 107 Drawing means (of the tree template generation system) 108 Point cloud data storage means (of the tree template generation system) 109 (Tree template storage means of the tree template generation system)

Claims

1. A system that generates a tree template that can be placed in three-dimensional space using a tree point cloud, which is a collection of three-dimensional measurement points obtained by measuring trees, A tree trunk model generation means generates a tree trunk model based on the tree trunk point cloud, which corresponds to the tree trunk among the tree point clouds, A branch and leaf model generation means generates a branch and leaf model based on the branch and leaf point cloud that corresponds to the branches and leaves of the tree among the tree point cloud, A missing area setting means for setting a missing branch / leaf region where the aforementioned branches and leaves are missing during measurement, A means for compensating for the missing branch and leaf region, which is used to complement the missing branch and leaf region with a part of the branch and leaf model, The system includes a tree template storage means for storing the tree template, which consists of the trunk model and the branch and leaf model in which the missing branch and leaf regions are complemented. A tree template generation system characterized by the following features.

2. A tree trunk diameter calculation means for calculating the tree trunk diameter of the aforementioned tree trunk model, The system further comprises a tree template acquisition means for reading another tree template having the same trunk diameter as the aforementioned tree template from the tree template storage means, The missing region completion means completes the missing branch and leaf region using a portion of the branch and leaf model that constitutes another tree template obtained by the tree template acquisition means. The tree template generation system according to claim 1.

3. The missing region completion means completes the missing branch and leaf region using a portion of the branch and leaf model that constitutes the same tree template. Furthermore, the missing region completion means completes the missing region of the branches and leaves by moving a part of the branch and leaf model so as to be symmetrical with respect to the trunk model, or by moving it so as to be point-symmetrical with respect to a reference point set on the trunk model. The tree template generation system according to claim 1.

4. The tree template storage means stores the branch and leaf model generated from the tree point cloud obtained by measurement from above, The tree trunk model generation means generates the tree trunk model using the tree point cloud obtained by measuring from the ground, The branch and leaf model generation means generates the branch and leaf model using the tree point cloud obtained by measurement from the ground, The missing region completion means reads the branch and leaf model related to the aerial measurement from the tree template storage means and completes the missing branch and leaf region with a part of the read branch and leaf model. The tree template generation system according to claim 1.

5. The missing region completion means enlarges or reduces a part of the branch and leaf model and then completes the missing branch and leaf region. A tree template generation system according to any one of claims 1 to 4.

Images