Digital Terrain Model Extraction Using Digital Aerial ... - IEEE Xplore

2012 IEEE 8th International Colloquium on Signal Processing and its Applications

Digital Terrain Model Extraction Using Digital Aerial Imagery of Unmanned Aerial Vehicle Wani Sofia Udin *1; Ahmad Farhan Hassan#2; Anuar Ahmad#2 & Khairul Nizam Tahar*3 *1

Faculty of Agro Industry & Natural Resources Universiti Malaysia Kelantan (Jeli Campus) Lock Beg No.100, 17600 Jeli, Kelantan, Malaysia e-mail: [email protected] #2

Department of Geoinformatics, Faculty of Geoinformation & Real Estate Universiti Teknologi Malaysia 81310 UTM Johor Bahru, Johor, Malaysia e-mail: [email protected]; [email protected] *3

Department of Surveying Science & Geomatics, Faculty of Architecture, Planning & Surveying Universiti Teknologi MARA 40450 Shah Alam, Selangor, MALAYSIA e-mail: [email protected]

and many more. A Digital Terrain Model (DTM) is a continuous representation of a ground surface landform [7]. In other words, DTM is described as the earth surface in the sense of the “bald earth” without human artifacts such as buildings, roads, bridges, and many more. The DTM is defined in different terms but similar meaning in some research [9], [5], [8]. DTM can be produced using many techniques or methods due to availability of the data.

Abstract— Digital Terrain Model (DTM) is extensively applied in various fields such as surveying and construction engineering, natural disaster management system, structure monitoring, and many more. Generally, the demand on accuracy and very detail information of DTM extraction is crucial for most of the applications. The advancement of photogrammetry and spin-off technology has advantageous in extracting highly accurate DTM. Currently, the DTM can be extracted from digital aerial imagery of small format camera mounted on light weight platform such as Unmanned Aerial Vehicle (UAV). The study is performs to assess the accuracy of DTM derived from UAV platform. Canon PENTAX W90 is utilized as non-metric camera for the earth information acquisition. In this study, 23 points of three dimensional coordinates (3D) were established using real time kinematic Global Positioning System (RTK-GPS) technique. Sixteen points are used in ground control point (GCP) for the purposes of aerial triangulation while seven as check point (CP) for accuracy assessment. The research output is then evaluated for planimetry and vertical accuracy using root mean square error (RMSE). Based on the analysis, sub-meter accuracy is obtained. As conclusion, UAV digital imagery can be used for accurate applications.

In photogrammetry, the DTM is derived from a series of stereo-image matching of digital aerial photograph. Commonly, the aerial photograph is captured using aircraft and metric camera. It is costly and time consuming due to appropriate planning in order to gain efficient information. Besides, it needs photogrammetry expertise to operate the camera during flying mode. However, aerial photogrammetry is expensive and improper technique especially for large scale mapping [1]. The advancement of photogrammetry and spinoff technology has advantageous in extracting highly accurate DTM. The compatibility of small format camera introduces new era for photogrammetrist. Furthermore, the widely used of Unmanned Aerial Vehicle (UAV) among civilian has beneficial photogrammetry world. The UAV is exploited as the-state-of-the-art platform for small format digital aerial imagery acquisition which reduces the problem of cloud cover. Nowadays, UAV system is also becomes very attractive among digital photogrammetric technique for various applications such as industrial, archaeology, architectural, geology, forestry, engineering and others. These missions are very dangerous to the pilot who

Keywords-Unmanned aerial vehicle, digital terrain model, digital camera, digital aerial imagery, accuracy

I.

INTRODUCTION

Digital Terrain Model (DTM) is extensively carried out in various disciplines such as surveying and construction engineering, disaster management system, structure monitoring

978-1-4673-0961-5/12/$31.00 ©2012 IEEE

272


conducts the flight and need require longer durations from one to another destination. Hence, the development of remotely controlled aircraft without human on board can be implemented. Based on the situations, this study is carried out to extract the DTM from digital aerial imagery of UAV platform. The DTM is then assessed with 3D coordinate established using Real Time Kinematic (RTK) Global Positioning System (GPS) technique. The study area is at Universiti Teknologi Malaysia (UTM) precinct by utilizing Canon PENTAX W90 digital camera mounted on Cropcam fixed-wing UAV.

II.

Figure 2. Canon PENTAX W90 digital camera

III.

RESEARCH METHODOLOGY

The research methodology adopted in this study is shown in Figure 3. Each phase of the study is explained as the procedure of DTM extraction using digital aerial imagery.

UNMANNED AERIAL VEHICLE (UAV) AND DIGITAL CAMERA

Unmanned Aerial Vehicle (UAV) is an aircraft, flying in the air with no pilot onboard and with capability for remote controlling the aircraft as describe by [3]. UAV was developed by military during World War I and II for reconnaissance and surveillance purposes [4]. Today, the civilian is able to operate the UAV for photogrammetry application. Equipped with GPS, INS, autopilot system and others, the system is designed to collect data for mapping and image interpretation purposes working on areas. The advantages in developing the technology of UAV for low altitude photogrammetric mapping are to perform aerial photography at cloudy day, to get full image of object under study from aerial, and to supply a cheap and easy system for high frequency needs of aerial photogrammetric survey [6]. Cropcam UAV is the small aircraft designed and considered as the type of low altitude UAV system as shown in Figure 1. It offers images on demand and is an inexpensive alternative to satellite or flying on the airplane over a field. Figure 3. The research methodology

A. Preparation Stage This stage investigates the purpose of the study. Due to focus under study, the photographic scale, flying height of UAV, coverage and others are determined. It involved the determination of 60% side lap and 30% end lap. A wellorganized image requires an essential arrangement because it is vital for data processing and analysis. Figure 1. Cropcam UAV

B. Planning Stage This phase involved the study area, software and instrument selection such as digital camera and GPS instrument. The digital aerial imagery is processed using ERDAS Orthobase 8.6 for producing DTM of UTM precinct. Unistrong E650 is used to establish the Ground Control Point (GCP) and Check Points (CP) for accuracy assessment.

Digital camera is categorized as non-metric camera where the camera is not specifically built for photogrammetric purposes. Digital camera is not characterized with fiducial mark, unstable calibration parameter, small format and many more [1]. The capability of small format camera is tested in this study for DTM extraction. The Canon PENTAX W90 digital camera is used for information acquisition. Figure 2 shows the Canon PENTAX W90 digital camera with interactive 12.1 Megapixel and 0.03 pixel size for high resolution small format camera.

C. Data Acquisition The digital aerial imagery is collected using digital camera mounted on Cropcam UAV. The Cropcam UAV is manually take-off at the proposed area as well as landing process.

273


Several strips of the image in JPEG (Joint Photographic Experts Group) are captured then transfer to the notebook for image processing. On-site calibration is also done for the aerial imagery.

IV.

RESULTS

Based on the study, three sets of output are obtained. The first result is camera calibration parameter derived from camera calibration process and second is DTM of the digital aerial imagery. The last outcome is orthophoto.

D. Establishment of GCP and CP The GCP and CP were established after the aerial photography mission and formation of uncontrolled mosaic is carried out. As for one strip of small format photograph 16 GCPs and 7 CPs were selected which enclosed the overlapped area as shown in Figure 4. For the establishment of GCP and CP, RTK GPS method was used.

A. Camera Calibration Parameters Table 1 shows the parameters of camera calibration parameters of Canon PENTAX W90 digital camera. The camera calibration parameters consist of the focal length (c), principal point offset (xp, yp), radial (k1, k2, k3) and tangential (p1, p2) lens distortion, “affinity” (b1) and different in scale factor (b2). TABLE I.

THE CAMERA CALIBRATION PARAMETERS OF CANON PENTAX W90

Parameter c(mm) xp(mm) yp(mm) k1 k2 k3 p1 p2 b1 b2

Figure 4. The location of GCP and CP

E. Camera Calibration Camera calibration is carried out before image processing is done. It is done by capturing convergence image of a test field which comprises of several targets and scale bar. The method used is self-calibration bundle adjustment method. After data acquisition of the test field, the images were processed using Australis software to obtain the camera calibration parameters.

Value 9.271 -0.123 0.118 -1.52e-04 2.37e-05 -1.68e-06 3.00e-04 -3.18e-04 4.54e-07 3.29e-04

Std Deviation 1.14e-02 6.38e-03 5.71e-03 1.36e-04 2.70e-05 1.70e-06 2.66e-05 2.49e-05 8.02e-05 8.50e-05

B. Orthopohoto Subsequently, a series of digital aerial imagery of the small format non-metric camera was used to produce orthophoto. The orthophoto was created after the formation of aerial triangulation of digital imagery with 60% endlap and 30% sidelap overlapping by using established GCPs. A stereomodel in three dimensional (3D) was successfully produced and orthophoto can be visualized. Figure 5 shows the orthophoto of UTM digital aerial imagery.

F. Data Processing Digital image processing is carried out in this phase for deriving DTM from UAV imagery. The ERDAS Orthobase is exploited to process the digital image of small format camera. The GCPs is used to perform the aerial triangulation in order to produce stereoscopic model. In the software, the 3D stereoscopic model was setup within short period of time [2]. The step is continued by generating DTM and orthophoto of the digital aerial imagery. DTM derived was used for accuracy assessment and orthophoto for visualization. G. Analysis Last stage of the study comprises of analysis in qualitative and quantitative technique. The qualitative is done by analyzing the quality of orthophoto and DTM generation. Meanwhile, the quantitative analysis is performed by using Root Mean Square Error (RMSE). The arbitrary coordinates system (X, Y, and Z) of control and check points derived using RTK GPS will be compared with estimates established by photogrammetry.

Figure 5. Orthophoto of UTM digital aerial imagery

274


C. Digital Terrain Model Figure 6 depicts the DTM generated based on aerial triangulation process. White color shows higher terrain elevation while grey displays the lower terrain elevation of UTM precinct. The quality of the DTM is then validates with accurate RTK-GPS of check points (CP).

CONCLUSION

Based on this study, the digital aerial imagery of UAV can be used for DTM extraction. The sub-meter accuracy produced by the data is relevant for various applications with low cost expenditure and less expert manpower. Besides, the flexibility and high efficiency of the UAV flight would be a solution for real-time mapping. It is because UAV can take-off and landing at limited open area with autopilot controlling. Lastly, the procedure of DTM extraction of UAV digital imagery can be implemented for various applications. ACKNOWLEDGMENT The authors wish to acknowledge use of the Australis selfcalibrating bundle adjustment program developed by Professor Clive Fraser at University of Melbourne, Australia. The authors would like to express their great thanks to Universiti Malaysia Kelantan (UMK) for supporting this work. Further, authors would like to thanks the private company known as JuruPro Sdn. Bhd. for consultancy and training. REFERENCES

Figure 6. DTM extracted from Cropcam UAV imagery

ANALYSIS

[1]

The DTM generated was analyzed by comparing with RTK GPS coordinate. The accuracy of DTM planimetry and vertical is shown in Table 2. In planimetry accuracy, a submeter ±0.555m and ±0.624 m were obtained for X and Y coordinates respectively. Meanwhile the RMSE for Z coordinates is ±1.117m. For average RMSE, ±0.766m was obtained by averaging the planimetry and vertical RMSE of small format digital imagery DTM.

[2]

TABLE II.

[3]

[4]

RMSE OF DIGITAL AERIAL IMAGERY DTM AND RTK GPS

[5]

[6]

[7]

[8]

[9]

275

Ahmad, A. Digital Mapping Using Low Altitude UAV. Pertanika Journal of Science and Technology. Vol. 19 (S): pg 51 – 58 Oct 2011. Ahmad, A., Hashim, K. A., Samad A. M. Aerial Mapping using High Resolution Digital Camera and Unmanned Aerial Vehicle for Geographical Information System. 2010 6th International Colloquium on Signal Processing & Its Applications (CSPA). Pg 201 – 206, 2010. Coppa, U., Guarnieri, A., Pirotti, F., and Vettore, A. Accuracy Enhancement of Unmanned Helicopter Poisitioning with Low Cost System, ISPRS (International Society of Photogrammetry and Remote Sensing. Vol. XXI Congress, Beijing, China, 2008. Eisenbeiss, H. A Mini Unmanned Aerial Vehicle (UAV): System Overview And Image Acquisition. International Workshop on Processing and Visualization Using High-Resolution Imagery. Nov 2004. Pitsanulok, Thailand, 2004. Haarbrink, R. B. and Eisenbeiss, H. Accurate DSM Production from Unmanned Helicopter Systems. The International Achieves of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII, Part B1, Beijing, 2008. Jian, L. Z. UAV for Mapping – Low Altitude Photogrammetric Survey. The International Archives of The Photogrammetry, Remote Sensing and Spatial Information Sciences. Part B1. XXXVII, 1183-1186, 2008. Podobnikar, T. Methods for Visual Quality Assessment of A Digital Terrain Model. S.A.P.I.EN.S 2.2. Institute of Photogrammetry and Remote Sensing, Vienna University of Technology, Vienna, Austria, 2009. Schumann, G., Matgen, P., Cutler, M. E. J., Black, A., Hoffmann, L., Pfister, L. Comparison of Remotely Sensed Water Stages From Lidar, Topographic Contours And SRTM. ISPRS Journal of Photogrammetry & Remote Sensing 63 (2008) 283–296., 2008. Yakar, M. Digital Elevation Model Generation by Robotic Total Station Instrument, Exp. Tech, 33 (2): 52-59, 2009