Abstract:
Capturing, documentation and visualization of natural heritage objects is an expanding application area. Advanced Unmanned Aerial Systems (UAS) offer a technology of obtaining geometrical and semantic information about natural landmarks from imaging using various types of sensors. The use of non-metric digital cameras embedded in middleclass UAS together with appropriate software and methodology allows for accurate digital spatial products to be obtained. These are files in standard formats with information that includes the orthophoto mosaic of the area under consideration, a three dimensional (3D) digital surface model (DTS), a 3D cloud, a vector CAD model (isolines and characteristic points), and a photo archive on the territory. The main focus of this study is on innovative lowcost photogrammetric technique for obtaining high-resolution datasets at a range of scales using the “Structure-from-Motion” (SfM). The SfM method is based on matching features (points) in multiple overlapping images in order to solve the camera pose and image geometry simultaneously and automatically. Image-processing creates 3D point clouds, originally in a local model space. Point clouds are transformed into a projection coordinate system using known camera positions or a network of GCPs on which high accuracy coordinates are measured by the GNSS (Global Navigation Satellite System) receivers and subsequently identified in the point cloud. The SfM method is useful when using non-metric cameras and is also applicable for convergent, horizontal, vertical and sloping images captured by different cameras. Example applications of SfM are presented for landforms of natural landmarks.