Forest resource investigation method based on laser radar and visible light synchronous photography technology

Abstract

The invention discloses a forest resource investigation method based on a laser radar and visible light synchronous photography technology, and belongs to the technical field of forestry. The method comprises the following steps: selecting a forest stand to be measured, and setting a scanning route; scanning the forest stand by using a laser radar and visible light synchronous photography technology; in the forest stand model, directly identifying tree species, and counting forest stand density; exporting point cloud data, performing individual tree segmentation according to a forest tree sequence, and generating DBH data and position coordinate data in batches; calculating the stand volume and tree species composition of the forest stand; determining a space structure of the forest stand according to the diameter at breast height data and the position coordinate data of the forest, and calculating a competition index of each tree; and linking the database with the forest stand model, establishing an augmented reality forest stand model, and realizing visualization of the forest resource data model. According to the method, the three-dimensional visual model of the forest stand can be established, the tree species can be directly identified, the forest resource survey data can be obtained through calculation, the survey efficiency is high, the cost is low, the data storage amount is small, and the later data processing difficulty is low.

Description

Technical Field

[0001] The invention relates to a forest resource survey method, in particular to a forest resource survey method based on laser radar and visible light synchronous photography technology, belonging to the technical field of forestry. Background Art

[0002] At present, the acquisition of forest resource survey data still mainly relies on personal experience for visual inspection or field measurement with simple tools. This traditional survey method is labor-intensive and inefficient. The data source will also cause errors due to the lack of a unified measurement standard. It can no longer meet the needs of modern forestry precision and information development. With the continuous development of measurement technology, high-precision measurement equipment and photogrammetry technology are also increasingly used in the forestry field.

[0003] Using the oblique photography technology of unmanned aerial vehicles, orthophotos and three-dimensional point cloud models of the survey area can be obtained. There are two main methods for obtaining tree height based on this technology. One is the point cloud segmentation method, which divides the three-dimensional point cloud into tree point cloud and ground point cloud, and obtains the tree height according to the difference between the tree top height and the average ground height; the other is the local maximum method, which first uses the three-dimensional point cloud to generate a digital surface model (Digital Surface Model, DSM) and a digital elevation model (Digital Elevation Model, DEM), and then performs grid difference between DSM and DEM to obtain a canopy height model (Canopy Height Model, CHM), and finally uses the local maximum method to filter the CHM to extract tree vertices and tree heights. The single tree heights obtained by these two methods have high accuracy, but these two methods mainly construct three-dimensional models of the forest canopy surface, and it is difficult to obtain real and effective point cloud information below the canopy and the ground layer due to the influence of forest canopy density, and the extraction of tree vertices will also be affected by the forest structure, resulting in omissions or multiple vertices of a tree. The single tree DBH is usually estimated by using the tree height, crown width, crown area and other parameters obtained from drone images and some ground-measured DBH to establish a regression model for inversion estimation, but this method has certain limitations in terms of tree species and regions. As the most advanced surveying and mapping method, 3D laser scanning technology has been widely used in forest resource surveys and forest stand structure research.

[0004] Using terrestrial laser scanning (TLS) and unmanned aerial vehicle laser scanning (ULS), high-quality three-dimensional models of forest stand spatial structure can be obtained, and more accurate spatial structural information such as DBH, tree height, and crown width can be extracted from them. In recent years, non-measuring devices such as digital cameras and smartphones have become the main tools for digital close-range photogrammetry, and some research results have been achieved in the measurement of forest stand parameters. The technology of integrating drone oblique photography technology with ground close-range photography technology to construct a refined three-dimensional model has been widely used in urban three-dimensional modeling, geological exploration, and architectural measurement. In forestry, laser radar and photogrammetry are combined for forest resource survey. Small surveying drones and smartphones are used for photogrammetry, and the spatial distribution characteristics of forest stands and the terrain characteristics under the forest can be more finely represented through image three-dimensional reconstruction. Wang Yi et al. used drone oblique photography to obtain the height information of the canopy surface of the sample plot, and the forest stand information was obtained using smartphone close-range photography. The drone oblique photography images and smartphone close-range photography images were reconstructed in three dimensions to obtain point cloud models, and the single tree height and DBH parameters with high accuracy were obtained.

[0005] In summary, the existing forest resource survey has the following shortcomings: 1. The traditional manual measurement survey method has a large workload and low efficiency, and the data source will also cause errors due to the lack of a unified measurement standard; 2. The height of a single tree obtained by the drone oblique photography technology has high accuracy, but it mainly constructs a three-dimensional model of the forest canopy surface, and it is difficult to obtain real and effective point cloud information below the canopy and the ground layer due to the influence of the forest canopy density. In addition, the extraction of tree vertices will also be affected by the forest structure, resulting in omissions or multiple tops for one tree, and the cost is high and difficult to promote; 3. The technology of integrating drone oblique photography technology and ground close-up photography technology to construct a refined three-dimensional model can more finely show the spatial distribution characteristics of the forest stand and the terrain characteristics of the understory, but there is no bark texture information of the trees, and the tree species cannot be identified, and thus the forest stand structure and tree species composition cannot be better determined; 4. The data storage volume is large, the later data processing is complex, and the workload is large. Summary of the invention

[0006] In order to overcome the shortcomings of the prior art, the present invention provides a forest resource survey method based on laser radar and visible light synchronous photography technology. The method obtains a three-dimensional visualization model of the forest stand, and the model contains the texture information of the trees, which can directly identify the tree species and determine the forest stand structure and tree species composition.

[0007] A forest resource survey method based on laser radar and visible light synchronous photography technology includes the following steps:

[0008] S1. Select the forest stand to be tested, divide it into several plots, set the scanning route, and require that each tree should not be scanned repeatedly when collecting data points;

[0009] S2. Scan the forest stand using laser radar and visible light synchronous photography technology, and set up an overlapping part including at least two trees at the junction of every two plots;

[0010] S3. In the forest stand model obtained after the forest stand scanning, the trees are numbered in order, and the tree species are identified according to the bark texture information of the trees, the tree numbers and tree species are entered into the database, and the forest stand density is counted;

[0011] S4, exporting point cloud data in the forest stand model, performing denoising, ground point classification and normalization on the point cloud data, segmenting the single trees according to the order of the trees, batch generating DBH data and position coordinate data of each tree, and adding the data to the database;

[0012] S5. Compare the one-dimensional timber volume table of the forest stand location and calculate the volume and tree species composition of the forest stand according to the tree species and DBH data in the database;

[0013] S6. Determine the spatial structure of the forest stand based on the DBH data and location coordinate data of the trees, calculate the competition index of each tree, sort the competition index, divide it into several levels in descending order, add the competition index and level to the database, and complete the forest resource survey.

[0014] This solution uses consumer-grade laser radar and visible light synchronous photography technology to establish a three-dimensional spatial visualization image model of the forest stand, obtain data information such as tree species, breast diameter, spatial position coordinates, etc. in the image model, and establish a forest stand attribute database through statistics and calculations to complete the forest resource survey. Then, the image model is superimposed with the attribute database to construct an augmented reality small class area three-dimensional spatial structure visualization image model to realize information visualization. On this basis, targeted planning and design of the forest stand can be carried out, implementation plans can be formulated, and production personnel can accurately implement them to accurately improve the quality of the forest stand.

[0015] Compared with the prior art, the present invention has the following beneficial effects:

[0016] 1. It saves more survey costs than manual survey, airborne lidar, ground-based radar, and even backpack radar.

[0017] 2. Scanning under the forest can clearly scan the trunks of the trees. The data acquisition accuracy is higher than that of airborne and ground-based radars. The accuracy of the DBH data extracted from the model is 95.24%. The relative distance between two adjacent trees is calculated using the position coordinate data extracted from the point cloud data and the measured data, with an accuracy of 84.14%.

[0018] 3. A three-dimensional visualization model of the forest stand can be obtained, and the model contains the texture information of the trees, which can directly identify the tree species and determine the forest stand structure and tree species composition. The accuracy rate of tree species identification in the model is 100% for artificial pure forests and mixed forests, and the accuracy rate of tree species identification in natural secondary forests is over 90%.

[0019] 4. Small data storage capacity. The forest resource survey method of the present application can be implemented in a 20m×20m sample plot, and the storage capacity of the model is only 20MB. Since the main trunk in the model is clear, it is relatively simple to perform single tree segmentation in the later stage, and the workload is greatly reduced.

[0020] The following is a further description of the scheme of the application in conjunction with the accompanying drawings and embodiments: BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 This is a flowchart of forest resource survey based on laser radar and visible light synchronous photography technology for this application;

[0022] Figure 2 It is an image model diagram of the experiment in the embodiment;

[0023] Figure 3 This is an example diagram of tree species that can be identified by bark texture information in a forest stand model in an embodiment;

[0024] Figure 4 This is a diagram showing the effect of single tree segmentation after point cloud data processing in the embodiment;

[0025] Figure 5 This is a data diagram obtained by splitting a single tree in the embodiment;

[0026] Figure 6 A comparison diagram of the actually measured diameter at breast height and the diameter at breast height extracted from point cloud data in the embodiment;

[0027] Figure 7 This is a comparison chart of the data of the relative distance between trees measured and extracted by the model in the embodiment;

[0028] Figure 8 It is a single tree information diagram displayed by the model after the augmented reality interactive link in the embodiment;

[0029] Fig. 9 A schematic diagram showing trees by tree species after augmented reality interactive linking in the embodiment;

[0030] Fig.10 This is a schematic diagram showing trees according to competition index levels after augmented reality interactive linking in an embodiment. DETAILED DESCRIPTION

[0031] The embodiments of the technical solution of the present invention will be described in detail below in conjunction with the accompanying drawings. Unless otherwise specified, the technical terms or scientific terms used in this application are generally understood by those skilled in the art.

[0032] Reference Figure 1 , a forest resource survey method based on laser radar and visible light synchronous photography technology includes the following steps:

[0033] S1. Select the forest stand to be tested and divide it into several small areas, each with an area of ​​0.5hm 2 , set the scanning route and require that each tree should not be scanned repeatedly when collecting data points;

[0034] S2. Scan the forest stand using laser radar and visible light synchronous photography technology, and set an overlapped part including at least two trees at the junction of every two plots, which is used for correction when the two plot models are spliced;

[0035] S3. In the forest stand model obtained after the forest stand scanning, the trees are numbered in order, the bark texture information of the trees can be obtained by visible light, the tree species are identified according to the bark texture information of the trees, the tree numbers and tree species are entered into the attribute database, and the forest stand density is counted;

[0036] S4, exporting point cloud data in the forest stand model, performing single tree segmentation according to the order of trees after denoising, ground point classification, digital elevation model generation and normalization processing of the point cloud data, batch generating DBH data and position coordinate data of each tree, and adding the data to the attribute database;

[0037] S5. Compare the one-dimensional timber volume table of the forest stand location and calculate the volume and tree species composition of the forest stand according to the tree species and DBH data in the attribute database;

[0038] S6. Determine the spatial structure of the forest stand based on the DBH data and location coordinate data of the trees, calculate the competition index of each tree, sort the competition index, divide it into several levels in descending order, add the competition index and level to the database, and complete the forest resource survey.

[0039] This implementation plan uses consumer-grade laser radar and visible light synchronous photography technology to establish a three-dimensional spatial visualization image model of the forest stand, obtain data information such as tree species, breast diameter, spatial position coordinates, etc. in the image model, and establish a forest stand attribute database through statistics and calculations to complete the forest resource survey. Then, the image model is superimposed with the attribute database to construct an augmented reality small class area three-dimensional spatial structure visualization image model to achieve information visualization. On this basis, targeted planning and design of the forest stand can be carried out, implementation plans can be formulated, and production personnel can accurately implement them to accurately improve the quality of the forest stand.

[0040] The present application scheme is further demonstrated below in conjunction with the embodiments:

[0041] 1. Select the forest stand to be tested. The forest stand to be tested was selected in the 33rd sub-class of the 80th forest class of Mengjiagang Forest Farm. A fixed standard plot was set up in the mixed forest of elm, mongolica and red pine with an area of ​​20m×20m. The diameter at breast height of each tree in the plot was measured with an accuracy of 0.1cm. The position coordinates of each tree were recorded with an accuracy of 0.1m.

[0042] 2. Laser radar and visible light synchronous photography. Install 3D scanning software on a smartphone with a laser radar. In this example, scaniverse software is used to shoot and scan a fixed standard area, obtain image information of all trees, and export a forest stand real-life model, such as Figure 2 As shown, the trunk scanning information of each tree is required to be more than 1.5m to ensure that the tree's breast diameter data can be obtained later.

[0043] 3. Tree species identification. The single tree in the 3D visualization model of the forest stand obtained after scanning is enlarged to obtain clear texture information of the tree bark. The tree species can be identified through the texture information of the bark, and the tree species are identified and recorded in sequence according to the tree number of the sample plot. According to the bark texture information in the model, three tree species, namely, Scots pine, ash and Korean pine, can be identified. That is, the experimental site is a mixed forest of Scots pine, ash and Korean pine. The identified tree species information is compared with the actual tree species information of the sample plot, with an accuracy of 100%. Figure 3 As shown, Figure 3 (a) Figure 3 (b) and Figure 3 (c) Represents the bark texture information of Pinus sylvestris, Fraxinus mandshurica and Pinus koraiensis respectively.

[0044] 4. Point cloud data preprocessing. Export the point cloud data file (*.lab) of the sample site in the model, and use the laser radar point cloud data processing software (LiDAR360 is used in this example) to denoise, classify ground points, and normalize the exported point cloud data file. Data denoising is an important step in the point cloud processing process, which can improve the efficiency of subsequent processing and the accuracy of data. Ground point classification is the basic operation of point cloud data processing. LiDAR360 uses the improved progressive encryption triangulation network filtering algorithm (Improved Progressive TIN Densification, IPTD) to classify ground points. First, a sparse triangulation network is generated through seed points, and then it is encrypted layer by layer through iterative processing until all ground points are classified. Normalization is the basis of forestry parameter extraction, which can remove the influence of terrain undulations on the elevation values ​​of point cloud data. Figure 4 The single tree segmentation map after preprocessing the point cloud data of the sample plot ( Figure 4 (a)) and the effect of dividing trees into different colors (as shown in Figure 4 (b)

[0045] 5. Data extraction. In the ground-based forestry module of LiDAR360, select the diameter at breast height measurement and the single tree segmentation based on the seed point, batch obtain the diameter data of the trees in the standard area at 1.3m of the tree orientation Z, and obtain the position coordinate data of each tree at the tree orientation X and tree orientation Y, such as Figure 5 shown.

[0046] Based on the data obtained in the above steps S1-S4, the following tests are performed:

[0047] 6. Accuracy test

[0048] (1) Accuracy test of DBH data. The DBH data extracted from the point cloud data were fitted with the measured data, with an accuracy of 95.24%. Three statistical indicators, total relative error (RS), average systematic error (E), and relative error absolute mean (RMA), were used for testing. The total relative error (RS) between the measured data and the point cloud data of the sample plot’s DBH was calculated to be 1.367, the average systematic error (E) was 1.194, and the relative error absolute mean (RMA) was 4.757. This shows that when the LiDAR360 software processes point cloud data with clear trunks, the accuracy of extracting DBH data is very high, such as Figure 6 shown.

[0049] (2) Accuracy test of position coordinate data. Since the position coordinate data extracted from the model adopts the 2000 National Geodetic Coordinate System, while the position coordinate data measured at the sample site is a temporary plane coordinate system, the two sets of data cannot be directly compared for accuracy test. However, both sets of data can calculate the relative distance between any two trees. Therefore, this study uses the relative distance calculated by the two sets of data for accuracy test.

[0050] The relative distance between two adjacent trees was calculated using the position coordinate data extracted from the point cloud data and the measured data, with an accuracy of 84.14%. The total relative error (RS) was calculated to be -0.475, the average system error (E) was -2.608, and the absolute mean error (RMA) was 15.85. The accuracy of the tree position coordinate data extracted from the point cloud data by LiDAR360 software is lower than that of the data extracted from the diameter at breast height. Figure 7 shown.

[0051] After the above-mentioned accuracy test, data analysis is performed on steps S5 and S6, and the data in the stand attribute database are statistically analyzed to calculate the stand density; the stand stock and tree species composition are calculated using the local one-dimensional timber volume table based on the tree species and breast diameter data; the single tree competition index is calculated using the breast diameter and tree position coordinate data, and the calculated data is added to the attribute database, thus completing the forest resource indicator survey of information such as stand species and composition, stand stock and stand density.

[0052] In order to achieve the visualization effect, an augmented reality interactive model is established. Specifically, the trees in the forest stand model are vectorized and linked to the database. When a single tree is clicked, the corresponding data information of the tree in the database is displayed, such as Figure 8 shown.

[0053] Display tree species information in the interactive model (refer to Fig. 9 ), display information by competition index level (refer to Fig.10 ), to achieve a more intuitive understanding of the basic information of forest stands and trees in the model, to carry out forest planning and design under visual conditions, and to establish a link between forest resource survey and attribute database. Figure 1 shown.

[0054] The present invention has been disclosed as above with preferred implementation cases, but it is not used to limit the present invention. Any technician familiar with the profession can make slight changes or modifications to equivalent implementation cases with equivalent changes by using the above-disclosed structures and technical contents without departing from the scope of the technical solution of the present invention, which still fall within the scope of the technical solution of the present invention.

Claims

1. A forest resource survey method based on laser radar and visible light synchronous photography technology, characterized in that: The method comprises the following steps: S1. Select the forest stand to be tested, divide it into several plots, set the scanning route, and require that each tree should not be scanned repeatedly when collecting data points; S2. Scan the forest stand using laser radar and visible light synchronous photography technology, and set up an overlapping part including at least two trees at the junction of every two plots; S3. In the forest stand model obtained after the forest stand scanning, the trees are numbered in order, and the tree species are identified according to the bark texture information of the trees, the tree numbers and tree species are entered into the database, and the forest stand density is counted; S4, exporting point cloud data in the forest stand model, performing denoising, ground point classification and normalization on the point cloud data, segmenting the single trees according to the order of the trees, batch generating DBH data and position coordinate data of each tree, and adding the data to the database; S5. Compare the one-dimensional timber volume table of the forest stand location and calculate the volume and tree species composition of the forest stand according to the tree species and DBH data in the database; S6. Determine the spatial structure of the forest stand based on the DBH data and location coordinate data of the trees, calculate the competition index of each tree, sort the competition index, divide it into several levels in descending order, add the competition index and level to the database, and complete the forest resource survey.

2. The forest resource survey method based on laser radar and visible light synchronous photography technology according to claim 1 is characterized by: In step S2, 3D scanning software is installed on a smartphone equipped with a laser radar, and the fixed standard area is photographed and scanned to obtain the image information of all trees, and the forest stand real-life model is exported. The trunk scanning information of each tree is required to exceed 1.5m to ensure that the tree breast diameter data can be obtained later.

3. The forest resource survey method based on laser radar and visible light synchronous photography technology according to claim 1 is characterized by: In step S3, the individual trees in the three-dimensional forest stand model are magnified to obtain the tree bark texture information.

4. The forest resource survey method based on laser radar and visible light synchronous photography technology according to claim 1 is characterized by: In step S4, the exported point cloud data file is denoised, ground point classified and normalized using LiDAR360 software.

5. The forest resource survey method based on laser radar and visible light synchronous photography technology according to claim 4 is characterized by: In step S4, the diameter at breast height measurement and the single tree segmentation based on seed points are selected in the ground-based forestry module of LiDAR360, and the diameter data of the trees in the standard land at the tree orientation Z are obtained in batches, and the position coordinate data of each tree at the tree orientation X and the tree orientation Y are obtained.

6. The forest resource survey method based on laser radar and visible light synchronous photography technology according to claim 1 is characterized by: The forest resource survey method also includes establishing an augmented reality interactive model, specifically: vectorizing the trees in the forest stand model and linking them to the database, so that when a single tree is clicked, the corresponding data information of the tree in the database is displayed.

7. The forest resource survey method based on laser radar and visible light synchronous photography technology according to claim 6 is characterized by: The data information includes tree species information and competition index level.

Images