Individual identification system for trees and timber

Abstract

We provide a low-cost and reliable wood traceability system. [Solution] The bark lenticels pattern of individual trees is naturally formed during the growth process and differs from tree species to tree and from tree to tree. Therefore, it provides unique information for each individual tree and cannot be easily falsified. By linking this bark lenticels pattern with traceability information and using it for the identification and matching of individual trees, a wood traceability system can be realized without attaching barcodes or electronic tags.

Description

Technical Field

[0001] The present invention relates to an identification system for an individual tree and an individual from which the wood originated in trees.

Background Art

[0002] Before the trees cut from the forest become final products such as buildings, they go through many processing and distribution stages. Therefore, it is difficult for suppliers and demanders to share information on the origin and subsequent history of the trees and wood used as materials. However, if raw material producers, processors, and consumers share such information, wood origin certification and quality assurance can be achieved, leading to improvements in distribution efficiency, inventory reduction, and processing yield. As shown in Non-Patent Document 1, enabling such information sharing is called wood traceability.

[0003] Wood traceability is implemented based on a database that holds information such as the origin, route, and processing of wood (hereinafter referred to as traceability information) and individual identification means. As individual identification means for trees and wood, attachments such as marks, barcodes, and RFID (electronic tags) have been used so far. Patent Document 1 and Patent Document 2 both propose using electronic tags for wood. Patent Document 1 aims to strictly manage the manufacturing process by holding the history data of wood processing in an electronic tag. On the other hand, Patent Document 2 aims to realize a system that provides traceability by holding information on the felling date, feller, tree species, origin, diameter, location information, and subsequent processing steps of a given tree individual in an electronic tag.

[0004] However, timber is often left outdoors not only while it exists as a tree, but also during its distribution process. The aforementioned tree identification methods have several problems, including the need for weather-resistant attachments, the risk of detachment, and the need to remove them beforehand if they interfere with processing. Furthermore, there is a risk of easily replacing, forging, and destroying the tags to misrepresent the origin of the timber. In addition, if an attempt is made to identify and manage individual trees over a long period before felling, there are problems such as the cost of attaching the tree identification methods and the fact that they become waste if lost. Moreover, electronic tags, in particular, incur purchase costs (Non-Patent Literature 1).

[0005] Furthermore, Patent Document 3 proposes a method for identifying wood by recording the pattern (tree rings) on the cut surface of felled logs. However, this method cannot be applied to the individual identification of trees before felling.

[0006] On the other hand, as an example of individual management other than for timber, Patent Document 4 proposes extracting features from facial images to perform person authentication for the purpose of identifying individuals, and Patent Document 5 proposes using the iris pattern of the eye to perform person authentication using a neural network. [Prior art documents] [Patent Documents]

[0007] [Patent Document 1] Japanese Patent Publication No. 2003-000000 [Patent Document 1] Patent No. 7273915 [Patent Document 2] Patent No. 7406796 [Patent Document 3] Patent No. 5538223 [Patent Document 4] Japanese Patent Application Publication No. 6-259534 [Patent Document 5] Patent No. 3855025 [Non-patent literature]

[0008] [Non-Patent Document 1] Journal of the Architectural Institute of Japan, Japan, Architectural Institute of Japan, October 2015, No. 21, No. 49, pp. 1143-1146 [Overview of the project] [Problems that the invention aims to solve]

[0009] The problem that this invention aims to solve is to provide a means for identifying individual trees in a forest over a long period of time, without using conventional methods for identifying individual trees such as barcodes or electronic tags, and for identifying the individual trees from which the timber originated during the distribution process from before logging to just before bark removal. [Means for solving the problem]

[0010] The present invention was made to solve the aforementioned problems, and is a tree individual identification system that utilizes individual differences in the bark lenticel pattern of individual trees, uses a database that stores bark lenticel pattern features and traceability information (including location information), and identifies individual trees by comparing them with bark lenticel pattern features of individual trees photographed at a different time than when individual identification was registered, or of felled logs. [Effects of the Invention]

[0011] According to the tree individual identification system of the present invention, individual trees can be identified as long as they have their bark attached, without the need to attach barcodes or electronic tags.

[0012] In addition, by including the location information of the trees (hereinafter referred to as the operation target area, such as pen, forest class, coordinates, etc.) in the traceability information and sharing this data in a certain area and using it as the basis for origin certification, the identity can be determined based on the origin information obtained by collation using the combination of tree bark texture pattern features and the declared origin information. Here, the tree bark texture pattern is formed by the growth process of each tree individual and cannot be easily disguised, so highly reliable origin certification can be provided.

[0013] In addition, by providing the current tree bark texture pattern features to the system during collation and adding them to the feature database, it can be updated to the latest state. With this feature, individual identification can be achieved even in the presence of disturbing factors such as tree growth and bark damage.

[0014] These features provide a means for identifying individual tree entities that form the basis of wood traceability. By applying this means of identifying individual tree entities to the wood circulation system, the circulation process can be made transparent, enhancing trust among raw material producers, processors, and consumers, and thereby facilitating smooth wood utilization.

Brief Description of the Drawings

[0015] [Figure 1] It is an example of the tree bark texture pattern used in the present invention. [Figure 2] It is a system configuration diagram in an embodiment of the present invention. [Figure 3] It is a diagram showing the photographing operation for obtaining the tree bark texture pattern in an embodiment of the present invention. [Figure 4] It is a flowchart for constructing a database in an embodiment of the present invention. [Figure 5] It is a diagram showing an example of how to obtain the tree bark texture pattern in an embodiment of the present invention. [Figure 6] It is a flowchart for collating an individual tree entity (before felling) in an embodiment of the present invention. [Figure 7] It is a flowchart for collating an individual tree entity (at the time of felling) in an embodiment of the present invention. [Figure 8] It is a collation flowchart of a tree individual (after felling) in an embodiment of the present invention. [Figure 9] It is a diagram showing a method for generating a bark pattern feature amount calculation model in an embodiment of the present invention.

Embodiments for Carrying Out the Invention

[0016] FIG. 1 illustrates the barks of several trees. The photos in the same column show the barks at different positions of the same individual. Also, the first and second columns are of the same tree species, and the others are of different tree species. Thus, the bark of a tree has different patterns depending on the tree species, individual, and position, and there is nothing that has exactly the same pattern. In the present invention, this property is utilized to identify an individual.

[0017] FIG. 2 shows the configuration of the system in this embodiment. The PC application (201) provides functions for registering, updating, and viewing traceability information to the user. The mobile application (202) provides functions for acquiring, collating bark pattern images on-site, viewing traceability information for each tree individual, and adding information. Registration, update, and acquisition of data from the PC application (201) and the mobile application (202) are performed by the server (203) via the Internet. The server (203) provides functions for calculating the feature amount of the bark pattern image input from the mobile application, input / output functions of data to / from the database (204), and a function for collating tree individuals.

[0018] Figure 3 illustrates the method for acquiring tree lenticel patterns in this embodiment. The user installs the mobile application provided by this system on their mobile device, uses its functions to capture images of the bark lenticel patterns, and transmits them to the system. Figure 3 top (301) shows the process of capturing images of the bark lenticel patterns of an unfelled tree. On the other hand, Figure 3 bottom (302) shows the process of capturing images of the bark lenticel patterns of a felled tree, i.e., a log. For bark lenticel pattern matching, the image is taken so that the direction of tree growth is at the top.

[0019] Figure 4 shows the flow for building a database for the traceability system in this embodiment. First, individual tree IDs are assigned to individual trees in the management area by some method (401). For each individual tree, the tree species, diameter at breast height, tree height, and individual location are identified, and traceability information for individual trees in the management area is obtained (403). The traceability information is registered and linked to the individual ID using the PC application (201) or mobile application (202) shown in Figure 2. In parallel with this, bark lenticel pattern images of each individual tree are acquired and input into the system (402). This series of processes is performed by the functions of the mobile application (202) shown in Figure 2. The system calculates feature quantities from the input bark lenticel pattern images and stores them in the database (405). The initial database (405) is constructed as the traceability information and feature quantities of individual trees present in the management area are registered in the database, linked by the individual ID.

[0020] Figure 5 shows a specific example of a method for inputting bark lenticel pattern images (402) as shown in Figure 4. To encourage inputting bark lenticel pattern images from the same location when matching individual trees, it is desirable to identify the shooting location using weather-resistant paint or the like (hereinafter referred to as the bark lenticel pattern image frame). In Figure 5, the four corners of the bark lenticel pattern image frame are marked with paint (501). Furthermore, by establishing rules for the height and direction of the shooting location, it is possible to estimate the shooting location even if the paint has deteriorated over time and is difficult to find. In addition, since the bark may be damaged for some reason, it is desirable to photograph multiple locations of the bark for a single individual tree. Figure 5 shows an example of setting the bark lenticel pattern image frame in the north-facing and east-facing directions at a height of 120 cm above the ground.

[0021] Figure 6 shows the flow for matching individual trees before felling. The user performs the matching by inputting bark lenticel pattern images of individual trees from a mobile application (602). At this time, the management area containing the target tree, the location of the tree, and the shooting direction (the direction the user's mobile device is facing) can be set and sent using the functions of the mobile application and used as matching hints (601). These hints allow the database to be narrowed down to target tree individuals and bark lenticel pattern images (603), reducing the number of feature sets to be compared and improving the efficiency of the matching. The input bark lenticel pattern images are converted into features (604) and compared with the narrowed-down set of features (605). If the degree of similarity is above a threshold as a result of the comparison (606), the information of the matched individual is displayed on the user's mobile application (607). At this time, the features derived from the input bark lenticel pattern images are newly saved as features of the target tree (609). If no individuals exceed the threshold for similarity, or if multiple individuals exist, the system will indicate "no match found" and prompt the user to re-enter the information (608).

[0022] Figure 7 shows a specific example of the matching flow during felling. This example uses attachments such as barcodes after felling, which contradicts the advantage of not using attachments, which is an advantage of the present invention. However, this does not diminish the significance of the present invention because it is limited to individual trees that have been felled and used, thus limiting costs, and the matching by bark strengthens individual identification using barcodes, etc.

[0023] Before felling individual trees, individual tree identification is performed (701). The identification operation at this time is the same as that shown in Figure 6, and is performed by comparing with a set of feature quantities stored in the database (707). If the identification of an individual tree is successful, a barcode to be attached to the individual tree is created (702). It is preferable to print the barcode on-site using a commercially available portable printer for mobile terminals, etc. The barcode contains traceability information such as individual ID, management area, and individual location, and provides a means to guarantee the traceability of individual trees by later matching this data with the bark lenticel pattern image. After felling and cutting into logs (703), in order to enable individual tree identification of the separated logs, a bark lenticel pattern image frame is added using weather-resistant paint (704), and the bark lenticel pattern image is input (705). The input bark lenticel pattern image has its feature quantities calculated on the server and is registered in the database (708). Next, the aforementioned barcode tag is attached to the cut surface of each log (706). The barcode tags attached to each separated log, the bark lenticel pattern image frames, and the features registered in the database make it possible to match each log to an individual tree.

[0024] Figure 8 shows a specific example of the flow for matching individual trees from felled and separated logs. After reading the barcode attached to the cut surface using a mobile application (801), the bark lenticel pattern image is input (802). The feature set of the individual tree that matches the individual ID contained in the barcode data is compared with the feature set derived from the input tree lenticel pattern image (803). If the degree of similarity is above a threshold (804), the user is shown that the match has been made (806). At this time, traceability information such as intermediate locations and processing details is added as needed (807). If it is below the threshold, the user is shown that the match has failed (805).

[0025] Figure 9 shows an example of a method for generating a model to calculate dendritic lenticel pattern features. The method shown here is a triplet network method, which is a type of deep distance learning. Three inputs are used: the original image (901) as the anchor, a transformed image (902) obtained by randomly applying rotation, translation, and color shift to the original image as a positive example, and another image (903) as a negative example. The CNN (904) then trains the convolutional neural network. In this example, the positive example (902) is the original image rotated. In this method, the loss function (905) is set so that the similarity between the original image and its transformed image is high, and the similarity between the original image and the other image is low. The degree of rotation, translation, and color shift, which are image transformations, is set based on the differences in images that may occur when images are actually compared. After training, dendritic lenticel pattern features are calculated by this model. [Industrial applicability]

[0026] By guaranteeing the traceability of timber, particularly proof of origin, we contribute to achieving a highly transparent timber distribution system.

Claims

1. A tree identification system that uses a database containing feature quantities of bark lenticel pattern and traceability information of individual trees, and identifies individual trees and obtains traceability information by comparing them with bark lenticel pattern feature quantities of individual trees or felled timber acquired at a different point in time.

2. A tree identification system characterized in that the traceability information according to claim 1 includes location information of individual trees, and the place (origin) from which the identified individual tree was collected can be confirmed even after felling.

3. The individual identification system according to claim 1, characterized in that the bark lenticel pattern can be newly acquired each time an individual tree is identified, thereby updating the feature quantities to reflect the latest state.

4. A timber distribution system equipped with a tree individual identification system according to any one of claims 1 to 3.

Images