Long line cliff topography measurement by using structure from motion algorithm in photogrammetry

Abstract

The topography of cliffs and steep slopes must be measured to acquire additional information for landscaping, visualizing changes and taking precautions against natural hazards. The Earth topography has been measured predominantly with photogrammetry, terrestrial/aerial laser scanning or other traditional measurement techniques. The stereo photogrammetry necessitates greater effort to obtain a three-dimensional (3D) model of the imaged surface. Meanwhile, terrestrial or aerial laser scanning can collect high-density measurements of spatial data in a short time. However, the costs of implementing laser scanning instruments are very high. Furthermore, conventional measurement techniques that use total stations require immense effort to collect complete 3D measurements of cliffs. On the other hand, dense image based point cloud using multi-view photogrammetry based on structure from motion (SfM) algorithm is much more effective than the others for measuring the Earth topography. In this study, the cliff topography of an old quarry located in the state of Selcuklu of Konya Province in Turkey was measured by multi-view photogrammetry. The cliff has a continuous length of approximately 600 metres and a height of 25 metres in some places. The 3D model of the cliff was generated with the image based dense point cloud of multi-view photogrammetry. Then 3D dense point cloud model was registered into a local georeference system by using control points (CPs). Because of the long line measurement area, number and localization of the CPs is very important for achieving a high-accuracy to registration into georeferenced system. The registration accuracies were evaluated for different number and distribution of the CPs with the residuals on the check points (ChPs). The high accuracy registration was acquired with uniform distributed 3 and 8 CPs as the residuals of 24.08 cm and 23.03 cm on the ChPs respectively. The results indicated that 3D measurement of long line cliffs can be performed using multi-view photogrammetry, and the registration should be made with the uniform distributed CPs. In addition, a texture-mapped 3D model and orthophoto images of the cliff surfaces were created for detailed visualization. © Authors 2019.