GENERATION AND ASSESSMENT OF HIGH RESOLUTION DIGITAL SURFACE MODEL BY USING UNMANNED AIR VEHICLE BASED MULTICOPTER
S. Gülci T. Dindaroğlu R. Gündoğan
AbstractGenerating digital surface models (DSMs) by unmanned air vehicles (UAV) and mountable systems become an appropriate and common method for scientific assessments and also for engineering related works. UAV classes, the specification of mounted sensors, flight height and speed may vary according to the aim and specific scope of the research (Watts et al., 2012; Wings et al., 2014; Gülci and Akay, 2016). The UAV-based studies that include before and after the flight stages, should be well designed for the quality of produced photogrammetric data and security (Akgül et al., 2016). This study examined a multicopter (hexacopter) as an air platform to seek opportunity in generating DSM with high resolution (Table 1). Flights were performed in Kahramanmaras Sutcu Imam University Campus area in Turkey. Preassessment of field works, mission, tests and installation were prepared by using a Laptop with an adaptive ground control station. Hand remote controller unit was also linked and activated during flight to interfere with emergency situations. Canon model IXSUS 160 was preferred as sensor. This sensor mounted on hexacopter has a record ability on the secure digital (SD) Card inside the camera, was mounted on air platform (Figure 1;2) (Remondino et al., 2011; Chao et al., 2016). Total of 8 ground control points were surveyed by using Global navigation satellite system (GNSS), which has almost millimeter accuracy in spatial measurements, and these points were considered as reference point in geo-rectification. Mission planner, which is an open-source interface software, provided flight mission to acquire block pattern. The flight altitude was defined 100 m, and the ratio of side and forward overlaps were planned as 80% (forward) and 60% (side). At the end of flight, total of 75 air photos were obtained from sensor (Figure 3). Processing and analysis of images were performed with PhotoScan, which works under the base of SfM (Structure from motion) approaches, and Cloud Compare, which is an open-source interface (Figure 4) (Westboy et al., 2012; Gülci et al., 2017; Cloud, 2017). Image processing steps by PhotoScan can be summarized as 1. identification of common points and creation of photo plane for block (alignment of photo), 2. point cloud generation, 3. image meshing, and 4. image texture (Agisoft, 2016). The initial options on image processing stages were implemented on PhotoScan as shown in Table 1. Hence, alignment of photo was completed with considering 74817 tie points detected. Then, analysis of dense cloud point generation figured out, totally, as 40.153.034 points (413.129 points/m2). Estimated image acquisition height was 111 m. The resultant resolution of the DSM and orthophoto were 4.92 cm and 2.46 cm/pix (Table 2). Briefly, topographic maps with high resolution can be derived from the use of UAV systems. It provides convenience for researchers by removing time and area constraints (Akgül et al., 2016). UAVs, which are presently defined as effective measuring instruments, can be used for measurements and evaluation studies in medium scale large fields. Accordingly, UAVs are effective tools that can produce high-precision and resolution data for use in geographic information system-based work. The orthophotos can be produced by RGB (Red-green-blue) images obtained with UAV , herewith information on terrain topography, land cover and soil erosion can be evaluated (d’Oleire-Oltmanns et al., 2012; Bending et al., 2014; Inan and Öztürk, 2016).