KITE-BORNE PHOTOGRAMMETRY FOR DECIMETRIC 3D MAPPING OF SEVERAL SQUARE KILOMETRES AREAS D. Feurer 1 , M. A. El Maaoui 2 , O. Planchon 1 , M. R. Boussema 2 , M. Pierrot-Deseilligny 3 1 IRD, UMR LISAH (INRA-IRD-SupAgro), F-34060 Montpellier, France, denis.feurer@ird.fr 2 Université de Tunis El Manar, ENIT, Laboratoire de Télédétection et SIRS, Tunisie 3 IGN/ENSG/DIAS et IGN/SR/LOEMI, Université Paris-Est, 94165 Saint-Mandé cedex, France ABSTRACT This paper presents a new methodology for the calculation of a digital elevation model (DEM) and the corresponding or- thophotography at very high resolution (20 cm for the DEM, 10 cm for the orthophotography) on a large surface (more than 3 km 2 ). Image acquisition was done with an off-the- shelf camera (Sony NEX-5N) attached to a delta kite. De- veloping the most accessible methodology being one of the goals of this study, the overall cost of the apparatus is limited to less than 1500 euros. This method has been developed to form an alternative with acquisitions from small Unmanned Aerial Vehicles (UAV). In particular, kite as payload carrier is seen as being complementary to lightweight multicopters, which are increasingly used for remote sensing applications [1, 2]. Kite is of a strong interest when the use of radio- controlled and motorized systems is hampered by local reg- ulation and/or when too strong winds keep these lightweight devices on ground. Index Terms— Aerial imagery, Digital elevation models, UAV, Kite, Photogrammetry, Structure from Motion, Frugal innovation, Micmac 1. INTRODUCTION With the development of UAVs and structure from motion (SfM) algorithms, very high resolution mapping has become accessible to most people. However, in a lot of countries, us- age of UAVs is strictly controlled and local regulations are the first hindrance to its development. As Colomina and Molina state, "Let them fly and they will create a market"[2]. Ad- ditionally, small and affordable UAV often suffer from a lack of autonomy and cannot be deployed under windy conditions, which limits their capabilities both in terms of coverage and responsiveness. In other words, current affordable technol- ogy does not cover exhaustively spatio-temporal scales and domains of very high resolution cartography. There is hence still a need for flexible and affordable very high resolution 3D mapping solutions of large areas. Limitations presented above led us to investigate the po- tential of a method based on kite-borne aerial imagery. Kite is historically one the the first platforms with hot-air balloons that have been use to acquire aerial imagery. Kite aerial pho- tography has even met a fairly good success in the field of archaeology (see for instance [3]) and has been used in geo- sciences (e.g [4]). However, existing literature does not takes full advantage of the recent advances of photogrammetric and SfM software, that now allow for the processing of big data sets. The aim of the work presented hereafter is hence to pro- pose a method that (i) refines existing principles of kite aerial photography and (ii) make good use of the huge power of re- cent SfM algorithms. Moreover, using affordable equipment, we thus propose a frugal innovation answer to the need of very high resolution mapping of several square kilometers areas. 2. MATERIAL AND METHODS 2.1. Material Criteria driving the material choice were cost, robustness, easy set up and in-flight reliability. For the platform, framed delta kite were used. They have been chosen within a large variety of kites because of their flight qualities (stability and high flight angles), easy mount with no need of adjustment, and fair payload. In this study, two delta kites, one of 4 m 2 and another one of 10 m 2 have been used. The line is a thin and light 200-lb Dyneema line. Low weight and thin kite line allows for better, more robust and constant flight characteristics. The line is graduated every 10 meters on the first 100 meters and then every 50 meters with a simple colour/thickness coding system. The rig is simply a tripod hung down a long line forming a pendulum. The long line ensures low-frequency movements of the rig around the ver- tical. The camera was chosen to meet a compromise between weight, image quality and geometry, and cost. The best com- promise found at the time of the experiment was the SONY NEX-5N, which allows to take 16Mpix images with fixed focal and disabled image stabilizer. A GentLED-Auto inter- vallometer has been used to trigger automatically the camera at given time intervals. Two autonomous QSTARZ BT1400S GPS logger were used, one attached onto the camera and the second on the kite operator. This positional information