edited summary tree diameter determination using digital ... · photomodeler scanner was used for...

Tree Age Estimation by Tree Diameter Measurement using Digital Close Range Photogrammetry (DCRP)

Muhd Safarudin Chek Mat*1, Mohammad Asyraf Mohd Nor *1, Jezan Md Diah#2, Mokhtar Azizi Mohd Din#3, Khairil Afendy Hashim*1 & Abd. Manan Samad*1

Pixelgrammetry & Al-Idrisi Research Group (Pi_ALiRG) Green Technology & Sustainable Development (GTSD), UiTM-RMI Communities of Research (CoRe)

*1 Centre of Studies Surveying Science and Geomatics

Faculty of Architecture, Planning and Surveying *2Faculty of Civil Engineering

Universiti Teknologi MARA Malaysia, Shah Alam, SELANGOR

#3Department of Civil Engineering

Faculty of Civil Engineering Universiti Malaya, Kuala Lumpur, WILAYAH PERSEKUTUAN

Email: [email protected]

Abstract - The digital close range photogrammetry (DCRP) approach has been used for tree’s age estimation from its photographs with image analysis. Age of a tree can be determined using diameter or from the circumference, then convert it into diameter of a tree. The aim of this study is to determine the age of a tree by measuring tree diameter using close range photogrammetry technique. Conventional method was used to estimate the age of a tree by determining diameter at breast height (DBH) by using a measuring tape. Three jelutong trees with a different size were used in this study. Finding of the results shows the 3D representation of three sample trees with their circumference and diameter measurement compared to the conventional reading. From the analysis it was found that the different between conventional and DCRP method on tree age estimation on tree id 020, 047 and 053 yields 0.313%, 0.850% and 0.424% respectively. Keywords - age tree estimation, the diameter of a tree, close range photogrammetry.

I. INTRODUCTION The tropical forests of Peninsular Malaysia are truly a

treasure trove of biodiversity. As the custodians of this priceless resource, we are responsible for its management and sustainable use or it will gradually diminish and vanish forever. Forest Research Institute Malaysia (FRIM) has been keeping many tree specimens to conserve and do the referencing for research purposes. Usually, FRIM will ask a freelance artist to do a sketch of the tree specimen and the sketching is not in scale. The specimen that has been stored also will change physically through time because of certain factors.

The jelutong (Dyera costulata) is a species of tree in the oleander subfamily. It lives in rainforests and its natural distribution is scattered locales in low-elevation tropical evergreen forest [1], [2] and [3]. Jelutong tree was used in this study based on the source of information and ready available at FRIM. Determination of the age of the trees can be done by measuring the diameter. A tree diameter commonly is measured about 1.37m above the ground. A steel diameter tape is commonly used to measure tree DBH and it must be calibrated to permit direct tree diameter readings [4].

Digital close-range photogrammetry (DCRP) is generally used in terrestrial photograph having object distance up to about 300m. The camera is usually accessible with terrestrial photograph, so that direct measurement can be made [5]. There have some advantages of photogrammetric procedures in a variety of applications using to other conventional methodologies because it more easy and fast result. The important one is the results produced can reach up to than millimeters accuracy in the final model [6] and [7].

In this study DCRP approach was applied in tree age estimation, thus introduce the other option to determine the age of the tree. By using close range photogrammetry technique, 3D representation of a tree can be created and could be stored digitally in its true dimensions and scale [6], [8] and [9]. For accuracy assessment, the diameter of tree from 3D representation can be measured and compared with the diameter of actual specimen while at the time the tolerance could be checked [8] and [10]. This research was conducted to perform and test the DCRP technique in comparison to the actual tree specimen.

2014 IEEE International Conference on Control System, Computing and Engineering, 28 - 30 November 2014, Penang, Malaysia

II. AIM AND OBJECTIVE The aim of this study is to determine the age of a tree by measuring tree diameter using close range photogrammetry technique. The objectives of this study are:

• To create tree 3D representation by using three samples as the references.

• To measure and compare the diameter of tree from 3D representations for the age of tree estimation.

• To estimate and compare the tree age from 3D representations with the referenced tree.

III. METHODOLOGY

The methodology of this study involves data acquisition, data processing, data presentation and data analysis. Firstly, the camera calibration and image capturing was performed. Then, the acquired images were exported to image processing software to create 3D model of the tree samples. Towards the end, the model was represented in digital mode and the result was analyzed by comparing with the conventional tree diameter measurement (Fig. 1). In this study, the images of three jelutong trees were obtained from FRIM. The identification numbers of the jelutong trees were recorded as 020, 047 and 053 which have different of ages. In common practice, the tree age can be determined by measuring its tree trunk’s diameter. However, [4] has described method to determine standing tree diameter by measuring 4 1/2 feet (1.37m) above the ground (foresters refer to this as the diameter at breast height (DBH) on the main stem. DBH normally measured to the nearest 1/10-inch and recorded by one-inch classes. For an example, trees with a DBH between 5.6 inches and 6.5 inches might be assigned as 6 inches when its information was recorded. For the implementation of the digital photographic works, a non-metric camera Canon EOS 1100D has been used in this study. While to create an accurate, high quality 3D models and measurements from photographs PhotoModeler Scanner was used for image processing and the procedure is known as photo-based 3D scanning. PhotoModeler Scanner is able to perform many modelling and measuring tasks, and differs from the basic PhotoModeler product by its ability to scan photo pairs to produce dense point cloud and meshed surfaces. The quality of the image product also depends on the type of camera. In specific the characteristic of the camera is as shown in TABLE 1.

TABLE 1. CHARACTERISTICS OF CANON EOS 1100D Feature Canon EOS 1100D

Resolution (pixel) 12.6 million Image size (pixel) 4272 x 2848

Size of Sensor (mm) 22.2 x 14.7 Auto focus Yes

Manual focus Yes Sensitivity ISO 100-6400

Fig. 1. Research methodology workflow.

A. Camera Calibration First, the non-metric camera needs to be calibrated. Camera Calibration is the process of finding the true parameters of the digital camera that took the photographs. Camera calibration is an important phase for non - metric camera as a necessary element for the assessment of photogrammetry. Fig. 2 shows the position of the camera during the image capturing in camera calibration.

Fig. 2. The position of the camera during camera calibration.


The purposes of camera calibration process are to

determine the focal length, principal point and distortion of the lens of the camera. This process is important to consider the unstable element in the camera as the internal orientation and lens distortion parameters.

Parameters need to be studies in processing digital image using DCRP approaches are focus length, lens distortion and main point. Focus length is a distance that perpendicular between the perspective system lens of camera and image plane camera. In this project focus length of the camera is 18.9838 mm. Lens distortion is happens where the camera lens elements are not parallel with the whole axis of the lens system. Besides, main point it is one position in camera that intersects between optical lens axis and image plane.

B. Image Capturing

The digital image of jelutong tree must be captured. The condition must be considered while taking a photograph is the distance, the angle, the situation of the environment and the way how to take the photograph. Camera station must be placed in appropriate. This is because the captured images will be processed in PhotoModeler Scanner where there are a few matters must be consistent with the rules of the PhotoModeler Scanner and the close-range principle. In order to avoid camera shaking, the camera is mounted on a tripod. Fig. 3 shows an example of on scene image capturing with codec target attached at tree. The codec targets are identified automatically by the program from the images. Coded target release by EOS System, which able to automatic detection and matching of point across multiple photographs. The high speed process of automatic referencing and orientation with less residual error now occur with this target [11].

Fig. 3. Example of digital photo taken on scene.

The position of the camera was placed at several appropriate distances to the position of the acquired images for clear and sharp visibility. In this study, the angles between the shots as was taken as close as 30°, with at least three photographs capture as a standard minimum

requirement to develop a 3D model. However, this study has conducted by establishing more than 10 camera station around the tree trunk (Fig. 4). This is to increase the overlap percentage between the adjacent photographs as much as possible and to ensure that the distance between two visible points or codec target in the scene can be seen clearly and could be used as reference scale for this project.

Fig. 4. Camera station position during image capturing.

C. Data Processing

At this stage the aims of this study to builds a 3D model of captured images. Basically, the processing stage needs a specific processing tool and in this case we had chosen Photomodeler Scanner running from Windows platform. The software is a window based program image processor which has specific module to extract 3D models and distance measurements from digital images. The software also allow users to capture more accurate details in a short time by using a camera as an input device. A 3D model is a set of connected 3D points, by which the three dimensional points have coordinate values for each of the Cartesian axes (X, Y, and Z). Fig. 5 shows a codec target that has been referenced.

Fig. 5. A codec target that has been referenced.

The points and the edges in the 3D model can be connected by patches, and they are called surfaces. These connections could help in visualizing the three dimensional


models when the models are projected onto a flat surface such as a computer monitor or a printed page (Fig 6 and 7).

Fig 6. A model that has been resurfaced.

Fig. 7. Digital measurement in PhotoModeler Scanner.

IV. RESULT AND ANALYSIS

A. Camera Calibration Parameters

The following TABLE 2 shows the result of camera calibration parameters. The camera calibrations consist of Focal Length (c), Format Size (w), Principal Point (X and Y), Lens Distortion (K1, K2, K3, P1 and P2).

TABLE 2. RESULT OF CAMERA CALIBRATION PARAMETER

Parameter Value Standard Deviation Focal length 18.983838 mm 0.003mm Format width 22.661702 mm 6.3e-004 mm

Principal point x (Xp) 11.316086 mm 0.002 mm Principal point y (Yp) 7.662923 mm 0.002 mm

Radial distortion 1 (K1) 5.266e-004 2.5e-006 Radial distortion 2 (K2) -1.248e-006 1.4e-008

Decentering distortion 1 (P1) -8.221e-005 1.7e-006 Decentering distortion 2 (P2) 6.958e-005 1.6e-006

B. Calculation of Diameter breast height (DBH) Diameter breast height (DBH) is the standard for

measuring trees. DBH refers to the tree diameter measured at 4.5 feet above the ground. DBH can be measured with a calibrated diameter tape, often referred to as a d-tape, which displays the diameter measurement when wrapped around the circumference of a tree. The circumference must be converted to diameter by solving for DBH in the equation: C = π * DBH Where: C = Circumference of tree, π = 3.142, DBH = Diameter Breast Height Therefore, DBH = C/ π TABLE 3 below shows the difference measurement of DBH from circumference between conventional method and digital close range photogrammetry (DCRP) method. Tree 020 has the smallest different follow by tree 053 and 047 with 0.312%, 0.425% and 0.849% respectively.

TABLE 3. CALCULATION OF DIAMETER BREAST HEIGHT (DBH) FROM CIRCUMFERENCE BY USING CONVENTIONAL

METHOD AND DCRP METHOD AND THEIR MEAN DIFFERENCE

Difference Tree ID

Conventional Method (cm)

DCRP Method

(cm) Value (cm)

(%)

020 43.285 43.420 0.135 0.312 047 33.100 33.381 0.281 0.849 053 50.859 51.075 0.216 0.425

C. Calculation of Diameter breast height (DBH) based on

Direct Measurement Non circular trees are to be measured in two

perpendicular diameters located as close as possible to the largest and the smallest diameter at that point, the average of these two is thus retained. Calculation of DBH based on Direct Measurement by Using DCRP Method:

Fig. 8. Non Circular Tree Measurement With Calliper


D = (d1 + d2) / 2 Where; d1 = The longest diameter d2 = The shortest diameter D = Diameter Breast Height (DBH)

TABLE 4 shows the difference measurement of DBH between conventional method and digital close range Photogrammetry (DCRP) method (direct measurement). Tree 053 has the smallest different follow by tree 047 and 020 with 0.641%, 0.743% and 0.896% respectively.

TABLE 4. MEASUREMENT OF DIAMETER BREAST HEIGHT (DBH) BETWEEN CONVENTIONAL METHOD AND DCRP METHOD

(DIRECT MEASUREMENT) AND THEIR MEAN DIFFERENCE

Difference Tree ID Conventional Method (cm)

DCRP Method (cm) Value (cm) (%)

020 43.285 42.897 0.388 0.896 047 33.100 33.346 0.246 0.743 053 50.859 50.533 0.326 0.641

D. Calculation the age of tree.

Determining the age of tree can be done by analyst the Diameter Breast Height (DBH) and Mean Annual Increment (M.A.I). To perform this analysis, simple mathematical formula is applied to obtain the estimation the age of the tree. A = DBH / M.A.I Where: A = Age of tree DBH = Diameter Breast Height M.A.I = Mean annual increment (1.13) Therefore, A = DBH/1.13 Jelutong tree mean annual increment is 1.13 cm. Based on the analysis between conventional and alternate method above, the observation show the reading of DBH based on the circumference of tree by using DCRP method are more closer to the reading of conventional method if compare to the reading of DBH from direct measurement. This is because of the sample of jelutong tree that has an irregular shape and cause the diameter measurement slightly different. So, the estimation of tree age based on DBH from circumference by using DCRP method were chosen to be compared with the estimation of tree age by using conventional method. TABLE 5 shows estimation of tree age by conventional and DCRP method, while TABLE 6 shows the comparison of tree age estimation.

TABLE 5. ESTIMATION OF TREE AGE BY CONVENTIONAL AND DCRP METHOD

Conventional Method DCRP Method Tree ID

DBH (cm) Age (year) DBH (cm) Age (year) 020 43.285 38.305 43.420 38.425 047 33.100 29.292 33.381 29.541 053 50.859 45.008 51.075 45.199

TABLE 6. THE COMPARISON OF TREE AGE ESTIMATION BETWEEN CONVENTIONAL AND DCRP METHOD

Age of Jelutong Tree Difference Tree ID Conventional

Method (year) DCRP Method

(year) Value (year)

(%)

020 38.305 38.425 0.120 0.313 047 29.292 29.541 0.249 0.850 053 45.008 45.199 0.191 0.424

V. CONCLUSION

This project shows that digital close-range photogrammetry (DCRP) was enabled to perform accurate measurement and potential for age estimation for three sample jelutong tree. Tree’s model generated from PhotoModeler Scanner was able to be measured and compared with conventional measurement. The results showed that comparison of tree age estimation between conventional and DCRP method for tree id 020, 047 and 053 are 0.313%, 0.850% and 0.424% respectively. The different is in small values. The beneficial from this study is a digital model of tree produced can be stored in database and able to use for future analysis. Besides, this technique also promises a tree data collecting without involves a lot of man power. Optimistically is hope that this study can contribute some knowledge for the related institution such as Forest Research Institute Malaysia (FRIM), Forestry Department and other related institutions.

VI. ACKNOWLEDGEMENT

Pixelgrammetry and Al-Idrisi Research Group (Pi_ALiRG); Green Technology & Sustainable Development (GTSD), UiTM-RMI Communities of Research (CoRe); UiTM Research and Management Institute (RMI-UiTM); Ministry of Higher Education (MOHE), FRGS RESEARCH GRANT [600-RMI / FRGS 5/3 (110/2012)]; Centre of Studies Surveying Science and Geomatics, Faculty of Architecture, Planning and Surveying, UiTM Shah Alam and Forest Research Institute Malaysia (FRIM) are greatly acknowledged.


REFERENCES

[1] Tropical Forest Biodiversity Centre (TFBC). (2004). Introduction. Retrieved from http://www.tfbc.frim.gov.my/welcome.html on 24 March 2013.

[2] Rainforest Relief. (2004). Jelutong (Dyera Costulata). Retrieved from http://www.rainforestrelief.org/What_to_Avoid_and_Alternatives/Rainforest_Wood/What_to_Avoid_What_to_Choose/By_Tree_Species/Tropical_Woods/J/Jelutong.html on 24 March 2013.

[3] Patrice. (2006). Dyera Costulata. Retrieved from http://commons.wikimedia.org/wiki/File:Dyera_costulata.jpg on 26 March 2013.

[4] Charles R. Blinn and Thomas E. Burk (2002). Estimating Tree Diameter. Retrieved from http://www.extension.umn.edu/distribution/naturalresources/components/dd3025-03.html on 24 March 2013.

[5] Wolf, P. R., & Dweitt, B.A. (2004). Element of Photogrammetry with Application in GIS (3rd edition ed.) : Mc Graw Hill.

[6] Pedro Arias, J. A., Javier Vallejo & Henrique Lorenzo (2009). Close-range Digital Photogrammety and Software Application Development for Planar Patterns Composition. Article, Department of Natural Resources and Environmental Engineering, University of Vigo, Spain and Superior School of Engineering. Bilbao, Spain.

[7] Mikhail, E.M., and Bethel, J. S. (2001). Introduction to Modern Photogrammetry. Purdue University: John Wiley and Sons.

[8] Abd Manan, S., Siti Zubaidah, B., Siti Aishah, A. R., and Khairil Affendy, H. (2009). Image Processing for Facade Mapping Using Digital Close Range Photogrammetric (DCRP) Approach, pp. 81-88.

[9] Zubair A. M., Zaimi M. and Mushairry M. (2006). The Use of Photogrammetry Technique to Evaluate the Construction Project Progress. Journal, Universiti Teknologi Malaysia.

[10] Ruinian Jiang A., Jauregui D.V., & White K.R. (2008). Close-range Photogrammetry Application in Bridge Measurement. Literature review, New Mexico State University, Department of Engineering Technology and Surveying.

[11] Muhd Safarudin, C.M., Jezan, M.D., Mokhtar Azizi, M.D., and Abd. Manan. S. (2014). Data Acquisition and Representation of Leaves using Digital Close Photogrammetry for Species Identification. Control and System Graduate Research Colloquium (ICSGRC), 2014 IEEE 5th, pp. 108-113.


edited summary tree diameter determination using digital ... · photomodeler scanner was used for...

Documents