Examination of images obtained from different unmanned air vehicles via automatic photogrammetric methods and accuracy analysis

Unmanned Aerial Vehicles (UAV) are the vehicles without pilots that carries payload with its own power system and is flown with automatic or remote control system. Even though the first examples of UAVs were arised with the 1st World War, UAV technology has made a big progress all around the world towards the end of 1990s. The UAVs, that are developed for the military areas mainly, are in great demand in civil use recently. This fast developing system has caused many different UAVs to arise. UAVs were classified according to their flight time, weight, elevation height and application areas. In order to generalize for this study, we can classify mini UAVs to two main sections according to their wing types. First one is with fixed wing UAV and the other one is multirotor UAV. Fixed wing UAVs are capable of flying long time, durable, have high payload capacity and suitable for vast areas. Multirotor UAVs have high maneuver capability and flight control and they are more likely suitable for use in small areas with their landing and take-off capabilities. These UAVs with different capabilities are commonly used in agriculture, forest, energy, mine, construction, archeology, architectural etc. areas as well as security, exploration, search and rescue areas. UAV systems, that became a new transporter platform with the developing technology, possessed an important place in cartography which prepares the basis of most of the engineering studies. Use of UAVs are seen revolutionary especially for photogrammetry. In classical measurement methods, the improvement of GPS (Global Positioning System) technology, provided measurements to be done more accurate and faster for photogrammetry just like in all other measurement methods. The integration of digital photogrammetry -which completed its formation recently- with air photogrammetry has provided a big advantage in vast areas. However the expensiveness of carrier platforms such as planes, helicopters and the difficulty of measurement repeat have showed up as the disadvantages of it. Therefore, it is continued to choose classical methods for measurements instead of air photgrammetry which is not frequently used. Today, ideal solutions were found with photogrammetric measurements done via UAVs for many areas. The main reasons for selection of this method are its low cost, speed, high accurency and capability of repetitive measurement. Thus, it is predicted that UAV photogrammetry will substitute for air photogrammetry. In this study, it is aimed to detect the usability of fixed wing UAV systems and multirotor UAV systems, which have different characteristics, for the same application area. The accuracy of the data obtained via these systems were analyzed by evaluating with automatic photogrammetric methods. In accordance with this purpose, `Silo Management" section of the project "Harvest Optimization in Beet Terrains in Boğazlıyan Region and Silo Management Pilot Project" which was conducted by Altoy Savunma Sanayi ve Havacılık A.Ş. and Kayseri Şeker Fabrikası A.Ş. in 2015 was benefited. For recording the volumetric change of beets that are waiting for processing, the data from 13th October 2015 which was obtained daily via Fixed Wing UAV and Multirotor UAV was used. The obtained images were processed with photogrammetric software support and examined with automatic photogrammetric methods. Under favour of predicted accuracy by literature research, 2 different UAVs were compared. First fixed wing UAV and multirotor UAV of advantages and disadvantages have been observed. Used in fixed wing Puma AV UAV most prominent feature of material resistant to difficult conditions due to the difference. Contains the secondary on the GPS and IMU's sensitivity is very high. 3.5-hour battery that causes the biggest trouble in time issue UAV this UAV is not included in. Large areas of flight, enabling data collection with ease PUMA UAV is a result of larger projects with efficiency. With Multirotor DJI phantom advanced UAV was found to be more appropriate to conduct studies in very small areas. It is a simple design are particularly easily bad weather UAV to be affected is the biggest disadvantage. However, multirotor will also be rare outside makes it from different angles. Digital elevation model, such as 3D modeling work in the production of multirotor advantage. In the studies done, fixed wing PUMA unmanned aerial vehicle and multicopter DJI phantom advanced unmanned aerial vehicle were used. While images were taken from 100 m with 24 MP camera of PUMA, DJI with 12 MP took images of the same area from 80 m and 300 m. In addition to creating the base surface to be used in volume calculations on October 10, 2015 at DJI with 80 m from the images. According to this, different UAVs were evaluated as well as the factors which the accuracy of data are dependent to were analyzed. Automatically it evaluated according to the fundamentals of photogrammetry images obtained with these UAVs. Among the most advantageous program for UAV photogrammetric software is Pix4d. Many studies in the literature have been performed with this software and the accuracy of the program has been proven. The images obtained with the two different type of UAV was evaluated by the same process steps. Firstly, ground control points and without ground control points, including the same processing steps were applied. Because adequate parameters found in the projects carried out as without ground control points, digital surface model and ortomosaic map was produced. Roughly calculated for positional accuracy of +/-3 m for DJI, +/-1.5m for Puma was found. Then the effect of ground control points to the final product was also investigated. During operation for the project with precision GPS ground control points received despite the preferred point of measurement (power pole, a sidewalk corner, etc.) has been seen but not healthy spots are marked as points apply. As a result, root mean square error of the calculated ground control points ; 1.81 cm YÖA with PUMA (100 m altitude) UAV's 0.079 m, 3.27 cm have YÖA DJI (80 m altitude) UAV 0.096 of m and have 12.95 cm YÖA DJI (300 m altitude) UAV's were 0.503 m. These results are evaluated considering the volume. Upon the choice of reliable and measurable points half a pixel or a pixel value is considered in error can be imported. Sample applications are provided silo pile in order to test the process steps to be followed in the volume calculation. EnsoMosaic & EspaCity software pile area to volume with sample is the calculated value exists, the Pix4d software downloaded data has been processed. The relative error in the use of volume measurements made by the base surface was found to be 2.60%. Automatically calculated with the volume of the base surface using triangulation is created through this method is approximately 13% has been a relative error. According to this assessment, ground control points used for volume projects account, primarily from the base surface data set October 10, 2015. Then, Digital Surface Models (DSM) with different methods of automatically generated point clouds was aimed at producing.In this context, the Delaunay Triangulation and Kriging interpolation methods are preferred. Produced DSM's vertical accuracy was tested by analysis. For vertical accuracy between the two methods of analysis, with 100 points, including a reference method KOH 's calculated and DJI-300 m in data is the highest standard deviation value of 0.045 m reached. Delaunay triangulation method of DSMs produced by PUMA reference data DJI 80m and DJI 300m silo pile of KOH 's position with 0.1828 m and 0.2366 m, respectively. Kriging interpolation method with reference to the data of PUMA DSMs produced by DJI80 and DJ 300m silo pile of KOH 's order is calculated by 0.1949 m and 0.2354 m. After the statistical comparison silo pile Pix4d, ground control point and without ground control point where images are processed as a result of the calculated volume value is automatically generated from the DSM and Puma produced with the lowest relative error of 3% GCP was reached volume. The result of the arithmetic mean is calculated as the volume of 1308.781 m³ silos pile. According to the absolute and relative error among other methods, the best volume 0.05% as a result of relative error of Kriging and Delaunay methods and data of Puma by calculating. Automatically produced with Pix4d DSM with a 1% relative volume made account in error. According to this study, if the program is automatically generated Pix4d making measurements required by DSM, Puma UAV SYM constituted features available depend on the minimum relative error can be provided with the data. Using the results of volume calculations using data from UAVs of different technical specifications, the most suitable method for agricultural studies has been analyzed. In the field of precision engineering volume accounts for what parameters is highlighted to be effective.