Assessment of Circle Fitting Method From Different Thicknesses and Heights On DBH Estimation Using Terrestrial Laser Scanner Data

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

2022 台灣地理資訊學會年會暨學術研討會

Assessment of Circle Fitting Method from Different Thicknesses


and Heights on DBH Estimation using Terrestrial Laser Scanner
Data
Gabriel Yedaya Immanuel Ryadi1, Michael Vashni Imamnuel Ryadi2, Chi-Kuei Wang3, Chao-
Hung Lin4

1,2,3,4 Department of Geomatics,


National Cheng Kung University, Tainan City 70101, Taiwan,
p68117018@gs.ncku,edu.tw, p68098098@gs.ncku.edu.tw, https://orcid.org/0000-0002-9058-
3902,https://orcid.org/0000-0001-8126-8794

Abstract:
A terrestrial laser scanner has great potential in forestry to assess the tree diameter with high
measurement accuracy. The tree’s diameter is one of the individual tree parameters in forestry that
describe how big a tree is but the value is immensely affected by its height from the ground. Thus,
DBH or diameter at breast height was chosen to be a standard way to express the diameter of the
tree by measuring the diameter at 4.3 feet or 1.3 m above the ground. The estimation of DBH in
this study utilized a circle fitting method by multi-height diameter and different cross-section
thicknesses of point clouds. Since these two diameter prediction methods are suggested in the
literature but there is no standard on elevation intervals for multi-height diameter or standard
cross-section thickness in terrestrial laser scanner application, this study evaluates the performance
of those two methods in DBH estimation with the same range elevation intervals and cross-section
thickness. Overall, those two approaches were statistically insignificant. According to the mean
absolute error (MAE), mean absolute percentage error (MAPE), root mean square error (RMSE),
and correlation coefficient (R2), the first assessment on 30 cm elevation interval achieved 3.23 cm,
7.58%, 4.11 cm and 0.971 while the second assessment on 30 cm thickness achieved 3.18 cm,
7.42%, 4.14 cm and 0.971. This study confirmed that in both approaches the multi-height diameter
and cross-section thickness of the point cloud is suitably used for DBH estimation. In addition,
this study suggests considering the number of data collection to get better tree measurement results.

Keywords: Diameter at breast height (DBH), circle fitting, point cloud, terrestrial laser scanner

1. Introduction
There are three parameters to assess the tree individually. One of those parameters is the
tree diameter. The tree diameter value is easily obtained by using diameter tape, but the value is
sensitive to its elevation from the ground (Koreň et al., 2020). Because of that, the diameter at
breast height (DBH) was chosen to express the tree diameter at 4.3 feet (1.3 meters) above the
ground (Hall et al., 1989; Sahin et al., 2019). Present-time, effective estimation of DBH is quite
important for forestry and ecological applications.
In the forestry field, the terrestrial laser scanner (TLS) has great potential to estimate the
tree diameter or DBH with high measurement accuracy by extracting the tree trunk point clouds
(Cabo et al., 2018; Witzmann et al., 2022). Two DBH estimation approaches based on the circle
fitting method were used in this study, including cross-section thickness-based (Koreň et al., 2020;
C. Liu et al., 2018; Xie et al., 2020) and multi-height diameter-based (G. Liu et al., 2018; L. Liu
et al., 2021).

1
2022 台灣地理資訊學會年會暨學術研討會

2. Subject
Regarding areas of study, most of the studies on DBH estimation utilize different
thicknesses or different elevation intervals. Starting from the thicknesses, the variation includes 2
cm thickness (Heo et al., 2019), 5 cm thickness (X. Wang et al., 2021; Zhou et al., 2019), 10 cm
thickness (P. Wang et al., 2019), 50 cm thickness (Xie et al., 2020), and 100 cm thickness (Koreň
et al., 2020). Then the variation of the elevation intervals in multi-height diameter includes 10 cm
elevation intervals (G. Liu et al., 2018) and 20 cm elevation intervals (L. Liu et al., 2021). Since
there is no elevation intervals standard or cross-section thickness standard for DBH estimation,
this study was conducted to evaluate the performance of DBH estimation in the same range of
cross-section thickness (10 to 40 cm) and with the same range of elevation interval (10 to 40 cm).

3. Methodology
3.1 Terrestrial laser scanner data acquisition
In this study, the point cloud data were collected along the sidewalk at National Cheng Kung
University, Cheng Kung Campus, Taiwan (22o59’54” N, 120o13’13” E). Figure 1a shows the
situation of the street, this street has 300 meters long with 12 meters in width. There were 28 trees
on the east side (Figure 1b) and 52 trees on the west side (Figure 1c).
This study used two instruments to collect the data. Those instruments are diameter tape and
Terrestrial Laser Scanner. The diameter tape has 2 meters long or 64π, this tape reads the tree
diameter directly in centimeters or inches units. The measurement from the diameter tape is the
reference data (recorded as the field measurement) for comparison and analysis. Then, the RIEGL
VZ-400i Terrestrial laser scanner was used to collect the point cloud data. This TLS has 5mm
accuracy and 3mm precision, with the high-speed data acquisition of up to 500.000 points/second.
This study used multiple station scanning to measure tree point clouds.

b.

a. c.
Figure 1. Study area. a. situation of the National Cheng Kung University. b. View of the east side
street. c. View of the west side street
3.2 Terrestrial laser scanner data processing
Point cloud data processing in this study includes pre-processing, point cloud segmentation,
and DBH estimation. The original point cloud was preprocessed using CloudCompare software.
Through CloudCompare software, the point cloud data was registered and the noise points were
removed. Still using CloudCompare software, the tree point clouds were segmented individually.
Next, the segmentation result was used for DBH estimation based on cross-section thickness and
based on multi-height diameters.

2
2022 台灣地理資訊學會年會暨學術研討會

3.2.1 DBH estimation


The circle fitting method is widely used for DBH estimation(Koreň et al., 2017; Witzmann
et al., 2022; Zeybek & Vatandaşlar, 2021). This method fits the circle to the sliced point at 1.3 m
elevation and projects into the 3D cartesian plane by rotating and translating the point cloud. The
formula of this method can be seen in Eq (1). The coordinates for the center of the circle denote as
(𝒙𝒄 , 𝒚𝒄 ), the radius as (𝒓), and the coordinates for the trunks point clouds denote as (𝒙, 𝒚).
(𝒙𝟐 − 𝒙𝒄 𝟐 ) + (𝒚𝟐 −𝒚𝒄 𝟐 ) = 𝒓𝟐 (1)
This study utilizes the circle fitting method with two approaches to estimate the DBH. Figure
2a and Figure 2b illustrate the two approaches for DBH estimation in this study. The first approach
(Figure 2a) is DBH estimation based on the cross-section thickness. This approach will consider
all the points to estimate one optimum DBH. (Koreň et al., 2020) concluded that there was an
increase in accuracy at different cross-section thicknesses. However, this method is difficult to
remove outliers that may come from irregular tree trunk shapes (bumps, cracks, or branches).
Then, the second approach (Figure 2b) is the DBH estimation based on the multiple-height
diameters. This approach will consider one DBH for each elevation interval, the mean value on Eq
(2) is the average of DBH estimations used to aggregate all values into one.
𝟏
̅̅̅̅̅̅
𝒅𝒃𝒉 = 𝒏 ∑𝒏𝒊=𝟏 𝒅𝒃𝒉𝒊 (2)
Furthermore, (L. Liu et al., 2021) conclude that the multi-height diameter approach in DBH
estimation selects DBH results by eliminating DBH values that are affected by outliers easily. In
this approach, multiple DBH are estimated over a range of point cloud slices across elevation
intervals. Meanwhile, the laser coverage affecting the DBH results in this approach, because more
point clouds have higher accuracy than the fewer ones.
a. b.

Figure 2. DBH estimation, the red circles are circle fitting results. a. Cross-section thickness
approach. b. Multi-height diameters approach.

3.2.2 Data analysis


The DBH estimations from TLS data were compared with the field measurements. The
mean absolute error (MAE), mean absolute percentage error (MAPE), root mean square error
(RMSE), and correlation coefficient (R2) were selected to evaluate the performance of DBH
estimations in this study. MAE measures the average absolute error between the DBH from TLS
measurements 𝒅𝒃𝒉𝒕𝒍𝒔𝒊 and DBH from the field measurements 𝒅𝒃𝒉𝒇𝒊 . MAPE measures the
accuracy level of 𝒅𝒃𝒉𝒕𝒍𝒔𝒊 compared to 𝒅𝒃𝒉𝒇𝒊 . RMSE measures the error rate of quadratic form
error value between the 𝒅𝒃𝒉𝒕𝒍𝒔𝒊 and the 𝒅𝒃𝒉𝒇𝒊 . Then the R2 measures the relationship between the
𝒅𝒃𝒉𝒕𝒍𝒔𝒊 and 𝒅𝒃𝒉𝒇𝒊 . The MAE, MAPE, RMSE, and R2 were calculated through Eq (2-5)

3
2022 台灣地理資訊學會年會暨學術研討會

𝒏
𝟏 (2)
𝑴𝑨𝑬 = ∑|𝒅𝒃𝒉𝒕𝒍𝒔𝒊 − 𝒅𝒃𝒉𝒇𝒊 |
𝒏
𝒊=𝟏
𝒏
𝟏 |𝒅𝒃𝒉𝒕𝒍𝒔𝒊 − 𝒅𝒃𝒉𝒇𝒊 | (3)
𝑴𝑨𝑷𝑬 = ∑ × 𝟏𝟎𝟎%
𝒏 𝒅𝒃𝒉𝒇𝒊
𝒊=𝟏
𝒏 𝟐
(4)
(𝒅𝒃𝒉𝒕𝒍𝒔𝒊 − 𝒅𝒃𝒉𝒇𝒊 )
𝑹𝑴𝑺𝑬 = √∑
𝒏
𝒊=𝟏
𝑪𝒐𝒗( 𝒅𝒃𝒉𝒇 , 𝒅𝒃𝒉𝒕𝒍𝒔 ) (5)
𝑹𝟐 =
𝝈𝒅𝒃𝒉𝒇 𝝈𝒅𝒃𝒉𝒕𝒍𝒔

4. Result and discussion


4.1 DBH Estimation

Figure 3. DBH estimation results from cross-section thickness and multi-height diameter
approaches
Figure 3 shows an example of the DBH estimation results from TLS measurements. DBH
estimation on the first approach (cross-section thickness) and the second approach (multi-height
diameter) in the 10 cm to 40 cm range. According to the figure, the DBH estimation from the first
approach slightly increases from 22.9 cm to 23.0 cm. Following that, the DBH estimation from
the second approach shows a consistent result which is 22.9 cm. Furthermore, the first approach
considers all the points to estimate one optimum circle fitting. Because of that, the trunk structure
such as branches, bumps, or textures may affect the DBH estimation. The noise removal step is
also important for this approach to achieve better DBH estimation. On other hand, the second
approach considers only several points at each interval to estimate DBH. Through this approach,
the DBH estimation that may be affected by trunk structure can be removed easily. However, the
multiple station scanning is necessary because the DBH estimation in the second approach depends
on the number of recorded point clouds on the trunks.

4
2022 台灣地理資訊學會年會暨學術研討會

4.2 Statistical Analysis


The statistical analysis in this study includes the MAE (mean absolute error), MAPE (mean
absolute percentage error), RMSE (root mean square error), and R2 (correlation coefficient) were
used to evaluate the performance of DBH estimation through cross-section thickness based and
multi-height diameter based. Table 1 shows the statistical result of DBH estimations from two
approaches compared with the field measurements. According to those results, there is no
significant difference between the DBH estimations from the cross-section thickness and the DBH
estimations from the multi-height diameter. Both of the approaches show the best range for cross-
section thickness and multi-height diameter in the 40 cm range.
Table 1. The results of MAE, MAPE, RMSE, and Correlation coefficient of DBH estimations

Methods MAE (cm) MAPE (%) RMSE (cm) R2

Cross-section Thickness
Range: 10 cm 3.39 7.74 4.39 0.968
20 cm 3.24 7.59 4.23 0.969
30 cm 3.23 7.58 4.11 0.971
40 cm 3.23 7.66 4.08 0.971
Multi-height Diameter
Range: 10 cm 3.25 7.59 4.25 0.969
20 cm 3.26 7.52 4.20 0.968
30 cm 3.18 7.42 4.14 0.971
40 cm 3.16 7.33 4.08 0.971

4.3 Conclusion
The use of terrestrial laser scanners in surveys has become more convenient and efficient for
obtaining 3D information. This study uses the TLS data to estimate the DBH using circle fitting
from two approaches. Based on the cross-section thickness and based on the multi-height diameter.
According to the results, there are no significant differences between those two approaches. This
study confirmed that in both approaches the multi-height diameter and cross-section thickness of
the point cloud is suitably used for DBH estimation. The method from this study needs to be
improved in future research, because the circle fitting method may be affected by the tree slopes.
In addition, this study suggests considering the amount of data collected to get better tree
measurement results (DBH estimation, tree height estimation, and tree crown area).

5. References
Cabo, C., Del Pozo, S., Rodríguez-Gonzálvez, P., Ordóñez, C., & González-Aguilera, D. (2018).
Comparing Terrestrial Laser Scanning (TLS) and Wearable Laser Scanning (WLS) for
Individual Tree Modeling at Plot Level. Remote Sensing, 10(4), 540.
https://doi.org/10.3390/rs10040540
Hall, R. J., Morton, R. T., & Nesby, R. N. (1989). A Comparison of Existing Models for DBH
Estimation from Large-scale Photos. The Forestry Chronicle, 65(2), 114–120.
https://doi.org/10.5558/tfc65114-2

5
2022 台灣地理資訊學會年會暨學術研討會

Heo, H. K., Lee, D. K., Park, J. H., & Thorne, J. H. (2019). Estimating the heights and diameters
at breast height of trees in an urban park and along a street using mobile LiDAR. Landscape
and Ecological Engineering, 15(3), 253–263. https://doi.org/10.1007/s11355-019-00379-
6
Koreň, M., Hunčaga, M., Chudá, J., Mokroš, M., & Surový, P. (2020). The Influence of Cross-
Section Thickness on Diameter at Breast Height Estimation from Point Cloud. ISPRS
International Journal of Geo-Information, 9(9), 495. https://doi.org/10.3390/ijgi9090495
Koreň, M., Mokroš, M., & Bucha, T. (2017). Accuracy of tree diameter estimation from terrestrial
laser scanning by circle-fitting methods. International Journal of Applied Earth
Observation and Geoinformation, 63, 122–128. https://doi.org/10.1016/j.jag.2017.07.015
Liu, C., Xing, Y., Duanmu, J., & Tian, X. (2018). Evaluating Different Methods for Estimating
Diameter at Breast Height from Terrestrial Laser Scanning. Remote Sensing, 10(4), 513.
https://doi.org/10.3390/rs10040513
Liu, G., Wang, J., Dong, P., Chen, Y., & Liu, Z. (2018). Estimating Individual Tree Height and
Diameter at Breast Height (DBH) from Terrestrial Laser Scanning (TLS) Data at Plot Level.
Forests, 9(7), 398. https://doi.org/10.3390/f9070398
Liu, L., Zhang, A., Xiao, S., Hu, S., He, N., Pang, H., Zhang, X., & Yang, S. (2021). Single Tree
Segmentation and Diameter at Breast Height Estimation With Mobile LiDAR. IEEE
Access, 9, 24314–24325. https://doi.org/10.1109/ACCESS.2021.3056877
Sahin, A., Kahriman, A., & Gokturk, A. (2019). Estimating diameter at breast height (DBH) from
diameter at stump height (DST) in triple mixed stands in the region of Artvin in Turkey.
Forestist, 69(1), 61–67. https://doi.org/10.26650/forestist.2019.18003
Wang, P., Gan, X., Zhang, Q., Bu, G., Li, L., Xu, X., Li, Y., Liu, Z., & Xiao, X. (2019). Analysis
of Parameters for the Accurate and Fast Estimation of Tree Diameter at Breast Height
Based on Simulated Point Cloud. Remote Sensing, 11(22), 2707.
https://doi.org/10.3390/rs11222707
Wang, X., Singh, A., Pervysheva, Y., Lamatungga, K. E., Murtinová, V., Mukarram, M., Zhu, Q.,
Song, K., Surový, P., & Mokroš, M. (2021). EVALUATION OF IPAD PRO 2020 LIDAR
FOR ESTIMATING TREE DIAMETERS IN URBAN FOREST. ISPRS Annals of the
Photogrammetry, Remote Sensing and Spatial Information Sciences, VIII-4/W1-2021,
105–110. https://doi.org/10.5194/isprs-annals-VIII-4-W1-2021-105-2021
Witzmann, S., Matitz, L., Gollob, C., Ritter, T., Kraßnitzer, R., Tockner, A., Stampfer, K., &
Nothdurft, A. (2022). Accuracy and Precision of Stem Cross-Section Modeling in 3D Point
Clouds from TLS and Caliper Measurements for Basal Area Estimation. Remote Sensing,
14(8), 1923. https://doi.org/10.3390/rs14081923
Xie, Y., Zhang, J., Chen, X., Pang, S., Zeng, H., & Shen, Z. (2020). Accuracy assessment and error
analysis for diameter at breast height measurement of trees obtained using a novel
backpack LiDAR system. Forest Ecosystems, 7(1), 33. https://doi.org/10.1186/s40663-
020-00237-0
Zeybek, M., & Vatandaşlar, C. (2021). An Automated Approach for Extracting Forest Inventory
Data from Individual Trees Using a Handheld Mobile Laser Scanner. Croatian Journal of
Forest Engineering, 42(3), 515–528. https://doi.org/10.5552/crojfe.2021.1096
Zhou, J., Zhou, G., Wei, H., Zhang, X., & Wang, X. (2019). Evaluation of Three Methods for
Estimating Diameter at Breast Height from Terrestrial Laser Scanning Data. IGARSS 2019
- 2019 IEEE International Geoscience and Remote Sensing Symposium, 6674–6677.
https://doi.org/10.1109/IGARSS.2019.8898005

You might also like