WO2022098139A1

WO2022098139A1 - System and method for identifying positions and number of trees using high-resolution drone image

Info

Publication number: WO2022098139A1
Application number: PCT/KR2021/015988
Authority: WO
Inventors: 장광민
Original assignee: 장광민
Priority date: 2020-11-06
Filing date: 2021-11-05
Publication date: 2022-05-12

Abstract

The present invention relates to a system and method for identifying the positions and number of trees using a high-resolution drone image, wherein information about the positions and number of trees can be automatically identified by applying GIS spatial analysis techniques on the basis of high-resolution drone image data acquired by drone aerial image capture.

Description

A system and method for identifying the position and number of trees using a high-resolution drone image

This application claims the benefit of priority based on Korean Patent Application No. 10-2020-0147817 dated November 06, 2020 and Korean Patent Application No. 10-2021-0100151 dated July 29, 2021, and All content disclosed in the literature is incorporated as a part of this specification.

The present invention relates to a system and identification method for identifying the position and number of a tree using a high-resolution drone image, and more specifically, to the position and It relates to a tree position and tree identification system and identification method using a high-resolution drone image that can automatically identify information about the number of trees.

In general, as a monitoring method to investigate the location and number of trees located in forested or non-forest areas, the direct survey method based on field surveys by manpower is predominant. R/S technology is being used as an auxiliary.

The conventional site survey method by manpower takes a lot of time and money for monitoring. In addition, if a large area is surveyed in a forest area with dense trees, an error may occur in the duplicate or omission of the number of standing trees. Accurate GPS reception is difficult under the crown layer of trees, so it is difficult to obtain accurate coordinates for the position of the trees. There are difficult limitations.

The use of remote techniques based on satellite and aerial images has the advantage of reducing the time and cost required for monitoring, but is mainly used to detect the canopy layer due to the limitation of low-resolution orthographic images. It is not used at all to identify a location or the like.

On the other hand, at the research level, methods for identifying individual trees by analyzing the shape and height structure of trees based on point cloud data obtained from LiDAR have been proposed and reported that high analysis accuracy can be secured. Due to the cost of filming lidar video, it is difficult to apply it to the field.

However, with the recent development of drone technology, it is possible to cost-effectively obtain high-resolution image data of 3 to 10 cm. All conditions are being prepared to develop and utilize it.

The present invention is a high-resolution drone that can automatically identify information on the position and number of trees by applying GIS spatial analysis technique based on high-resolution drone image data acquired by drone aerial photography as derived to solve the above-mentioned problems. It is intended to provide a system and method for identifying the position and number of trees using an image.

A system and identification method for identifying the position and number of trees using a high-resolution drone image according to an embodiment of the present invention is a point cloud data (Point cloud ), a first digital space that generates first digital spatial information data through a digital surface model (DSM) and a digital elevation model (DEM) from the aerial image data. An information generating unit, a second digital spatial information generating unit generating second digital spatial information through a Normalized Difference Vegetation Index (NDVI) from the spectral image data, Excluding the height value data of an artificial structure from the point cloud data And a first Modified CHM (Modified Canopy Height Model) unit for storing the standing tree height value data, the first Modified CHM unit through the tree height value data stored in the tree top to identify the tree top (Tree top) and a detection unit, wherein the first modified CHM unit collects and processes the first digital spatial information data and the second digital spatial information data.

In one embodiment, the first digital spatial information generation unit applies the numerical surface model (DSM) to apply the numerical surface model unit to derive the height values of the trees and artificial structures, the numerical surface model (DEM) to apply the numerical elevation model (DEM) A numerical elevation model unit for deriving the height value of the terrain excluding the height value of the artificial structure, and a water pipe height model unit for applying the water pipe height model by collecting the data values derived from the numerical surface model unit and the numerical elevation model unit. can be characterized as

In an embodiment, the second digital spatial information generating unit derives the regular vegetation index (NDVI), and the regular vegetation index (NDVI) calculates the difference between the values of the near infrared and red light bands between the near infrared and the red light band. It may be characterized in that the calculation is divided by the sum of the values.

In an embodiment, the first modified CHM unit may be characterized in that the first digital spatial information data is removed from the second digital spatial information data to derive the tree height data value.

In one embodiment, the head detection unit Local Maxima for extracting the maximum height value for each predetermined interval from the height data value derived from the first Modified CHM unit, the Standing height derived from the first Modified CHM unit. Concave area for extracting the area of the convex part at regular intervals from the data value, the second Modified CHM part for extracting the standing tree having a height greater than or equal to the reference value from the tree height data value derived from the first Modified CHM part, and the Local It may be characterized in that it includes a data collection unit that collects the values derived from the Maxima unit, the Concave area unit, and the second Modified CHM unit to generate information on the position and number of trees.

The method for identifying the position and number of trees using a high-resolution drone image according to another embodiment of the present invention is performed by the point cloud data generator, and the aerial image data and the spectral image data captured from the drone equipped with the spectral sensor are matched to the photographic image. The step of generating point cloud data, which is performed by the first digital spatial information generator, is performed from the aerial image data through a digital surface model (DSM) and a digital elevation model (DEM). generating first digital spatial information data, the second digital spatial information generating unit generating second digital spatial information from the spectral image data through a Normalized Difference Vegetation Index (NDVI); The first Modified CHM (Modified Canopy Height Model) unit is performed, the step of excluding the height value data of the artificial structure from the point cloud data, and storing the tree height value data, the first Modified CHM unit is performed in the first head detection unit and identifying a tree top of a tree through the tree height value data stored in the , wherein the storing of the tree height value data includes the first digital spatial information data and the second digital spatial information It may be characterized by collecting and processing data.

In an embodiment, the step of generating the first digital spatial information is performed by a numerical surface model unit, and applying the numerical surface model (DSM) to derive height values of the trees and artificial structures, the numerical elevation model unit The step of applying the numerical elevation model (DEM) to deriving the height values of the terrain excluding the height values of the trees and artificial structures and performed in the canopy height model part, the numerical surface model part and the numerical elevation model It may be characterized in that it comprises the step of applying the crown height model by collecting the data values derived from the part.

In an embodiment, the generating of the second digital spatial information may include: deriving the NDVI through the second digital spatial information generating unit; and calculating the normal vegetation index (NDVI) by dividing a difference between the near-infrared light and the red light band by the sum of the near-infrared light and the red light band.

In an embodiment, the storing of the standing tree height value data may include deriving the standing tree height data value by removing the second digital geospatial data data from the first digital geospatial data data. there is.

In one embodiment, the step of identifying the tree top of the tree is performed in the Local Maxima part, and from the tree height data value derived from the first Modified CHM part, the maximum height value for each predetermined interval is extracted. step, performed in the concave area part, extracting the area of the convex part at regular intervals from the tree height data value derived from the first Modified CHM part, is performed in the second Modified CHM part, and the first Modified The step of extracting a tree having a height greater than or equal to the reference value from the tree height data value derived from the CHM part and the data collection part are performed by collecting the values derived from the Local Maxima part, the Concave area part, and the second Modified CHM part. It may be characterized in that it comprises the step of generating information on the position and the number of trees.

According to one aspect of the present invention, since it is possible to automatically acquire information on the location and number of trees in forested areas and non-forest areas, it can be used for systematic lumber resource management. When combined with the canopy height model (CHM), it is possible to estimate the exact height (height), so it can be used to cost-effectively monitor the height and fraction, which are important parameters necessary for estimating the amount of tree stacking or carbon dioxide absorption. have an advantage

1 is a diagram illustrating a process of identifying a standing tree position and a tree number using the standing tree position and number identification system 100 using a high-resolution drone image according to an embodiment of the present invention.

FIG. 2 is a diagram illustrating topography according to data values through Digital Surface Model (DSM), Digital Elevation Model (DEM), Normalized Difference Vegetation Index (NDVI), and Canopy Height Model (CHM).

3 is a diagram illustrating a model according to height data values of trees and artificial structures.

4 is a diagram illustrating a process of detecting a tree top of a standing tree according to the standing tree position and tree number identification system 100 using a high-resolution drone image.

5 is a diagram illustrating a state in which a tree is identified and detected according to the tree position and tree number identification system 100 using a high-resolution drone image.

Hereinafter, preferred examples are presented to help the understanding of the present invention. However, the following examples are only provided for easier understanding of the present invention, and the content of the present invention is not limited by the examples.

Referring to FIG. 1 , the tree position and number identification system 100 using a high-resolution drone image includes a point cloud data generation unit 110 , a first digital spatial information generation unit 120 , and a second digital spatial information generation unit 130 . , is configured to include a first Modified CHM unit 140 and a super head detection unit 150 .

First, the point cloud data generating unit 110 is to generate a point cloud data by registering the aerial image data 111 and the spectral image data 112 photographed from a drone (not shown) equipped with a spectral sensor to a photographic image. can

The first digital spatial information generation unit 120 generates first digital spatial information data from the aerial image data 111 through a digital surface model (DSM) and a digital elevation model (DEM). can

Here, by using the numerical elevation model and the numerical elevation model, a canopy height model that can estimate the height of vegetation and artificial structures from the ground surface can be created. Therefore, when artificial structures and trees are mixed, such as vegetation in downtown, a post-processing process is required to artificially remove height information of artificial structures in order to extract tree location information.

Looking at the first digital spatial information generation unit 120, the first digital spatial information generation unit 120 applies a numerical surface model (DSM) to derive the height values of the trees and artificial structures (DSM unit) , 121), the numerical elevation model part (DEM part, 122) and the numerical surface model part 122 and the numerical elevation that derive the height values of the terrain excluding the height values of the trees and artificial structures by applying the numerical elevation model (DEM) It may include a water pipe height model unit 123 that applies the water pipe height model by collecting data values derived from the model unit.

On the other hand, in the case of using a spectral image taken by a drone equipped with a spectral sensor, it is possible to easily produce NDVI image data useful for reading vegetation areas through a combination of spectral image bands.

In more detail, the regular vegetation index is an index for indexing the growth status of crops using the reflectivity of near-infrared rays. The value of is expressed as a quantitative value. In general, the vegetation area represents a positive (+) value, and the non-vegetated area is expressed as a value less than or equal to 0, and the index can be used to distinguish between the vegetation area and the non-vegetation area where artificial structures are located.

In this regard, the second digital spatial information generating unit 130 may generate the second digital spatial information from the spectral image data through a regular expression index (NDVI), and the regular expression index (NDVI) is a combination of near-infrared rays and near-infrared rays as in the above equation. It can be calculated by dividing the difference in the values between the red light bands by the sum of the near infrared and red light bands.

Derived a Modified CHM model in which height information of vegetation areas is extracted through spatial analysis processing that removes altitude values located in non-vegetated areas using regular vegetation index (NDVI) images from the existing canopy height model. can do.

In more detail, the first modified CHM unit 140 includes the second digital spatial information data generation unit 130 derived from the first digital spatial information data generation unit 120 from the first digital spatial information data generation unit 120 . The digital spatial information data may be collected and processed, and more specifically, the tree height data value may be derived by removing the second digital spatial information data from the first digital spatial information data.

Using the modified CHM data, it is possible to extract the point where the height values at two different scales (analysis units) have the local maxima, and the designated scale unit is the object tree. In order not to miss detection, it can be analyzed by applying a value smaller than the minimum planting interval of the target area. Considering that even if trees are densely planted in Korea, it is common to plant 3,000 trees in an area of 1 ha. ) can be derived.

In addition, when the upper part of the crown layer of the tree is viewed from above at a right angle, it has an upwardly convex geometric shape. Therefore, if the concave area is extracted at the 1.5 m scale in consideration of the general planting interval in Korea using the revised canopy height model data, it may be possible to detect the place expected to be the upper part of the tree crown.

Next, in order to classify trees or shrubs in the revised canopy height model data, a height of 5 m or more can be applied as a threshold to identify standing trees. When it is necessary to irradiate young trees of 5 m or less, including seedlings, or to include shrubs or glue trees with a height of 3 to 5 m, it can be applied by lowering the standard value to 5 m or less or not applying it.

If applied to the present invention using the above contents, the super head detection unit 150 may be configured to include a Local Maxima unit 151 , a Concave area unit 152 , and a second Modified CHM unit.

The Local Maxima unit 151 extracts the maximum height value for each predetermined interval using the tree height data value (modified tree crown height model data) derived from the first Modified CHM unit 140 to determine the height value of the highest standing tree. can be derived

The Concave Area unit 152 may extract the area of the convex portion at regular intervals by using the tree height data value (modified crown height model data) derived from the first Modified CHM unit 140 .

The second Modified CHM unit 153 may extract a standing tree having a height equal to or greater than a reference value by using the tree height data value (modified crown height model data) derived from the first Modified CHM unit 140 .

Through the above process, a point that satisfies all the criteria for the regional maximum value, the upper crown area, and the height of 5 m is selected as a candidate group for individual tree identification. By selecting the tree, the coordinate information of the object tree and the number information are generated. Accordingly, the data collection unit 150 of the present invention uses the tree height data value (modified tree crown height model data) derived from the first Modified CHM unit 140 to the Local Maxima unit 151, Concave Area unit ( 152) and the second Modified CHM unit 153 may generate information on the location and number of trees by collecting each data derived through the unit.

Referring to FIG. 2, FIG. 2 is a view showing some of the images collected by using a drone equipped with aerial imaging equipment and a spectral image sensor in the area of 70-5 Hwadong-ri.

It is a diagram showing the spectral images collected through the drone using a numerical elevation model (DSM, 121) and a numerical elevation model (DEM, 122). The regular vegetation index (NDVI, 130) is used to be useful in reading the vegetation area through the spectral image band combination as described above. In addition, the regular vegetation index is an index for indexing the growth status of crops using the reflectivity of near-infrared rays. It is expressed as a quantitative value. In general, the vegetation area represents a positive (+) value, and the non-vegetation area is expressed as a value of 0 or less.

In addition, if the DEM (122) image using the numerical elevation model is removed from the DSM (121) image using the numerical elevation model, an image using the Canopy Height Model (CHM, Canopy Height Model, 123) model can be obtained. A modified canopy height model in which the height information of the vegetation area is extracted after the spatial analysis process of removing the altitude value located in the non-vegetated area using the regular vegetation index (NDVI, 130) image from the crown height model (CHM, 123) (Modified CHM) can be derived.

A drawing 122-1 in the upper left of FIG. 3 is a diagram illustrating a model in which the elevation value of the terrain is stored numerically using a digital elevation model (DEM), excluding the height values of trees and artificial structures. The drawing 121-1 in the upper right is a model in which the elevation values of the ground including the height values of trees and artificial structures are stored numerically using DSM (Digital Surface Model), and the drawing 121-1 in the upper right is As shown in , it can be confirmed that the height values of the ground surface and artificial structures are measured and stored together with the height values of trees. In addition, the lower left drawing 123-1 is the height of trees and artificial structures using CHM (Canopy Height Model), which removes the upper left drawing 122-1 from the upper right drawing 121-1. It is a diagram showing a model in which values are stored numerically, and the drawing 140-1 in the lower right is a model in which the height values of trees and artificial structures are stored numerically using the CHM of the drawing 123-1 in the lower left. It is a diagram showing the deriving of a modified canopy height model (Modified CHM) from which height information of vegetation areas is extracted through spatial analysis processing that removes altitude values located in non-vegetated areas using regular vegetation index (NDVI) images. .

Referring to FIG. 4 , the superhead detection unit 150 may include a Local Maxima unit 151 , a Concave area unit 152 , and a second Modified CHM unit 153 . The Local Maxima unit 151 can extract the maximum height value for each predetermined interval from the tree height data value derived through the first Modified, and the height value in scale 1 and scale 2 is local. It can be confirmed that the point with the maximum value is extracted. The green arrow bar indicates that the point having the locally maximum height value among scale 1 is extracted, and the yellow arrow bar indicates that the point having the maximum height value locally on scale 2 is extracted. say what happened Here, when the interval between the yellow and green arrows is 1.5 m or less, the maximum value of the height extracted from scale 1 and scale 2 is compared and the higher one is stored, and the lower one can be deleted.

The concave area unit 152 may extract the area of the convex part for each predetermined interval from the tree height data value derived from the first modified CHM unit. Also in the concave area, it is possible to extract an area with a convex shape at a scale interval of generally 1.5 m.

In addition, the second Modified CHM unit 153 may classify trees 5 m or more tall in order to distinguish trees and shrubs from the image derived from the first Modified CHM unit 140 , and 5 m or more as a threshold value. can be applied to identify the standing tree. At this time, it may be possible to investigate young trees of 5 m or less, including seedlings, or by lowering the standard value to 5 m or less or not applying it if investigation is necessary including shrubs or glue trees with a height of 3 to 5 m.

The data collection unit 154 may collect data that satisfies all the conditions in the Local Maxima unit 151 , the Conacave area unit 152 , and the second Modified CHM unit 153 and identify them as an entity. Here, when the distance between the identified trees is adjacent to within 1 m, a point having a high effort may be selected as the individual tree.

Referring to FIG. 5 , a diagram 151-1 in the upper left is a diagram illustrating the detection of superhead candidates by applying the Local Maxima algorithm. The upper right drawing 151-2 is a view of detecting the upper part of the canopy layer, the lower left drawing 151-3 is a view showing a combination of the left and right upper drawings, and the lower right drawing 151-4 is In the lower left drawing 151-4, candidate groups of adjacent individual trees within 1 m can be selected and merged. Here, selecting and merging candidate groups of adjacent individuals within 1 m is to prevent overestimation.

As a method of identifying the position and number of trees using a high-resolution drone image, it is performed in the point cloud data generation unit, and the aerial image data and the spectral image data taken from the drone equipped with a spectral sensor are matched to the photographic image to generate the point cloud data. can be generated, and it is performed by the first digital spatial information generating unit, and the first digital spatial information data from aerial image data through digital surface model (DSM, Digital Surface Model) and digital elevation model (DEM, Digital Elevation Model) can create In addition, the second digital spatial information generating unit generates second digital spatial information through a Normalized Difference Vegetation Index (NDVI) from the spectral image data, and the first Modified CHM (Modified Canopy Height Model) unit It is performed in , and it is possible to save the height value data of the artificial structure by excluding the height value data of artificial structures from the point cloud data. (Tree top) can be identified.

Here, the step of generating the first digital spatial information is performed in the numerical surface model unit, and by applying the numerical surface model (DSM), the height values of the trees and artificial structures can be derived, and the numerical elevation model unit is performed. , by applying the numerical elevation model (DEM), it is possible to derive the height values of the terrain excluding the height values of the trees and artificial structures, and the data derived from the numerical surface model part and the numerical elevation model part are performed in the canopy height model part. By collecting the values, the canopy height model can be applied.

Also, in the generating of the second digital spatial information, a regular vegetation index (NDVI) may be derived through the second digital spatial information generating unit, and the difference between the near-infrared and red light bands is a value obtained by adding the near-infrared and red light bands. By dividing by , the regular expression index (NDVI) can be calculated.

In addition, the storing of the tree height value data may include removing the second digital spatial data data from the first digital spatial data data to derive the tree height data value.

Finally, the step of identifying the tree top of the standing tree is performed in the Local Maxima part, and from the tree height data value derived from the first Modified CHM part, the maximum height value for each predetermined interval can be extracted, and Concave It is performed in the area part, and from the height data value derived from the first Modified CHM part, the area of the convex part at regular intervals can be extracted, and it is performed in the second Modified CHM part, From the tree height data value, it is possible to extract a tree having a height greater than or equal to the standard value, and it is performed in the data collection unit. information can be created.

Although the above has been described with reference to preferred embodiments of the present invention, those skilled in the art can variously modify and change the present invention within the scope without departing from the spirit and scope of the present invention as set forth in the claims below. You will understand that it can be done.

Claims

A point cloud data generation unit for generating point cloud data by matching aerial image data and spectral image data taken from a drone equipped with a spectral sensor to a photographic image;

a first digital spatial information generator for generating first digital spatial information data from the aerial image data through a digital surface model (DSM) and a digital elevation model (DEM);

a second digital spatial information generator for generating second digital spatial information from the spectroscopic image data through a Normalized Difference Vegetation Index (NDVI);

A first Modified CHM (Modified Canopy Height Model) unit for storing the height value data of the trees except for the height value data of the artificial structures in the point cloud data;

A tree top detection unit for identifying a tree top of a tree through the tree height value data stored in the first Modified CHM unit; including,

The first Modified CHM unit,

The first digital spatial information data and the second digital spatial information data are collected and processed.

A system for identifying the position of the tree and the number of trees using a high-resolution drone image.
According to claim 1,

The first digital spatial information generating unit,

a numerical surface model unit for deriving height values of trees and artificial structures by applying the numerical surface model (DSM);

a numerical elevation model unit that applies the numerical elevation model (DEM) to derive the height values of the terrain excluding the height values of the trees and artificial structures; and

It characterized in that it comprises;

A system for identifying the position of the tree and the number of trees using a high-resolution drone image.
According to claim 1,

The second digital spatial information generating unit,

The regular vegetation index (NDVI) is derived, and the regular vegetation index (NDVI) is calculated by dividing the difference between the values of the near-infrared and red light bands by the sum of the near-infrared and the red light bands,

A system for identifying the position of the tree and the number of trees using a high-resolution drone image.
According to claim 1,

The first Modified CHM unit,

It is characterized in that by removing the second digital spatial information data from the first digital spatial information data to derive the height data value,

A system for identifying the position of the tree and the number of trees using a high-resolution drone image.
According to claim 1,

The super head detection unit,

a Local Maxima unit for extracting a maximum height value for each predetermined interval from the tree height data value derived from the first Modified CHM unit;

Concave area for extracting a convex area for each predetermined interval from the height data value derived from the first Modified CHM unit;

a second Modified CHM unit for extracting a tree having a height greater than or equal to a reference value from the height data value derived from the first Modified CHM unit; and

A data collection unit generating information on the position and number of trees by collecting the values derived from the Local Maxima unit, the Concave area unit and the second Modified CHM unit; characterized in that it comprises a;

A system for identifying the position of the tree and the number of trees using a high-resolution drone image.
It is performed by the point cloud data generation unit, generating a point cloud data (Point cloud) by registering the aerial image data and the spectral image data photographed from the drone equipped with a spectral sensor to the photographic image;

generating first digital spatial information data from the aerial image data through a digital surface model (DSM) and a digital elevation model (DEM), performed by the first digital spatial information generating unit;

generating second digital spatial information from the spectral image data through a Normalized Difference Vegetation Index (NDVI), performed by the second digital spatial information generator;

The first Modified CHM (Modified Canopy Height Model) unit is performed, the step of excluding the height value data of the artificial structure from the point cloud data and storing the height value data of the tree;

The step of identifying the tree top of the tree through the tree height value data stored in the first Modified CHM unit and performed by the tree head detection unit;

The step of storing the tree height value data is,

Collecting and processing the first digital spatial information data and the second digital spatial information data;

A method of identifying the position and number of trees using high-resolution drone images.
7. The method of claim 6,

The step of generating the first digital spatial information comprises:

It is performed in the numerical surface model unit, applying the numerical surface model (DSM) to derive the height value of the tree and artificial structures;

It is performed in the numerical elevation model unit, applying the numerical elevation model (DEM) to derive the height value of the terrain except for the height values of the trees and artificial structures; and

It is performed by the water pipe height model unit, and applying the water pipe height model by collecting the data values derived from the numerical surface model unit and the numerical elevation model unit; characterized in that it comprises,

A method of identifying the position and number of trees using high-resolution drone images.
7. The method of claim 6,

The generating of the second digital spatial information comprises:

deriving the regular vegetation index (NDVI) through the second digital spatial information generator; and calculating the regular vegetation index (NDVI) by dividing the difference between the near-infrared light and the red light band by the sum of the near-infrared light and the red light band.

A method of identifying the position and number of trees using high-resolution drone images.
The method of claim 1,

The step of storing the tree height value data is,

Deriving the height data value by removing the second digital spatial information data from the first digital spatial information data; characterized in that it comprises,

A method of identifying the position and number of trees using high-resolution drone images.
7. The method of claim 6,

The step of identifying the tree top of the stand,

It is performed in the Local Maxima unit, extracting a maximum height value for each predetermined interval from the height data value derived from the first Modified CHM unit;

It is performed in the concave area part, extracting the area of the convex part at regular intervals from the tree height data value derived from the first Modified CHM part;

It is performed by the second Modified CHM unit, extracting a tree having a height greater than or equal to a reference value from the tree height data value derived from the first Modified CHM unit; and

It is performed by the data collection unit, and generating information about the position and number of trees by collecting the values derived from the Local Maxima unit, the Concave area unit, and the second Modified CHM unit; characterized in that it comprises;

A method of identifying the position and number of trees using high-resolution drone images.