An unmanned aerial vehicle (UAV), commonly known as a drone and also referred to as an unpiloted aerial vehicle or a remotely-piloted aircraft (RPA) by the International Civil Aviation Organization (ICAO) is an aircraft without a human pilot aboard. Entrepreneurship together with robotic, computer vision, and geomantic technologies have established a new paradigm of aerial remote sensing and mapping that, for some years now, has been serving the needs of large-scale low-altitude imaging and geospatial information users and therefore, has developed the industry of its own. In UAV-based photogrammetry, a data sensor is mounted on a remote-controlled drone that receives data from a low altitude. It has several applications in different areas due to high flexibility, low cost, ease of access and suitable accuracy. Recent improvements in sensors and aerial platforms have broadened its applications ranging from volume calculation, orthophoto generation and 3D reconstruction, to large-scale map generation [1–3].
Producing large-scale maps from highlands (1/1000 and larger) is one of the main requirements in surveying projects, which has been addressed by traditional field mapping using a total station, Global Navigation Satellite System (GNSS) receivers, and levels in recent years. Such maps are used in road construction, excavation volume calculation, and dam construction. The necessity to collect dense and precise points, besides the complexity of the regions has always turned this work to one of the hardest tasks in surveying. However, the advent of UAV-based photogrammetry has made such works much easier. network design management is the key to competitive exploitation of UAS for photogrammetry and remote sensing [1]. For mapping, if the terrain is flat or almost flat, the usual method of capturing the terrain with a UAV is to fly horizontally at a constant height above the mean sea level (MSL). In the abruptly changing terrain, the flight altitude must adapt to the ground’s height in each flight line instead of maintaining a constant height above the MSL. The reason for latter is because, when a UAV is flying at a constant height above the MSL, researchers found out that the vertical root mean square error (RMSE) values were larger in areas with complex topography compared within flat areas [4, 5]. In complex topography, the distance between the sensor and the ground is not constant, and the overlap is reduced and could become critically low in very steep areas, which causes fewer images to overlap in steeper areas compared with low areas [6]. Network design management are the key to a competitive exploitation of UAS for photogrammetry and remote sensing [1]. The most important part of network design in UAV-based photogrammetry projects is flight planning. Flight design is completely dependent on the scale of the map, the topography of the area, and the maneuverability of the flight platform.
In UAV-based photogrammetric projects, having stability in ground sample distance (GSD) results in uniformity of the scale in the photogrammetric blocks and improves the geometry of the project and increases the altitude accuracy, especially for orthophoto generation. Data acquisition from mountainous, and semi-mountainous areas with a classic two-dimensional flight without considering the actual topography of the area in the flight planning causes severe differences in scale in the photogrammetric block, potential accident of the drone. Moreover, UAV images typically affected by radiometric distortions generated by different factors such as sunlight variations, sensor calibration and topographic conditions [7–9]. Another problem occurred in such cases is capturing blurred images [10] due to the camera focus, which is considered to be fixed when flying over a flat surface in a 2D network design. Solving these problems requires having a fixed flight altitude over the ground based on a digital elevation model (DEM) as the true terrain representation, which reduces the influence of varying altitude above ground level (AGL) distances and the subsequent distortions and inaccuracies on the final results.
Currently, software solutions are more and more available for planning flights in UAV-based photogrammetric projects. Despite this, available solutions rely on low-resolution and low-accuracy global terrain models (SRTM). However, such low-resolution terrain models (generally with GSD higher than 30 m) have average elevation inaccuracies about ± 10 m, and up to ± 30 m in mountain areas [11]. Thus, they are not suitable for applications in high mountainous environments as it has enormous effect on the planned AGL consequently on the specified elevation accuracy [12].
In this paper, a new imaging network design algorithm for UAVs based on a digital elevation model is presented. The main objective of this work is to plan a flight for a UAV in a 3D environment, keeping a constant flight level regarding the ground, that is to say, a terrain-following flight (TFF). The algorithm also changes inter-frame overlaps based on the height difference of the region to either reduce the data redundancy, cost, and time of the project or to avoid the gap and missing data.