HIGH DYNAMIC RANGE IMAGES IN SAND DUNE PHOTOGRAMMETRY

We conducted a study of sand dunes using High Dynamic Range (HDR) images as input for close-range digital photogrammetry. Photogrammetry matches points between images taken from two or more perspectives, then uses parallax between the matched points to create a 3D model of the scene. Previous photogrammetric studies highlighted several barriers to creating models accurate enough to determine sand dune movement over short time periods. The biggest obstacle is that dune surfaces appear homogenous at distances of more than a few meters, yielding poor point matching. Vegetation movement that inhibits point matching, foreground vegetation that obscures parts of the scene, and low viewing angles from which to photograph many dune surfaces also cause difficulty in modeling.

HDR images can reveal more details from uniform subjects because they are derived from three normal pictures taken in rapid succession, as underexposed (-2 stops), overexposed (+2 stops), and normal exposures. The camera must be mounted firmly to a tripod to avoid movement while it captures the three images. The images are combined and processed using tone mapping techniques to create single HDR images that photogrammetry software then uses to make 3D models of the scene. A 3D model using HDR images is more detailed and uniform than one from normal exposures alone, with a 3D digital terrain model that is more dense and shows more details and variation in the surface. The HDR images help to correct for differences in lighting as photography progresses during data gathering and therefore the model’s surface color is more uniform. However, movement of vegetation can cause a ghosting effect in the finished photos. Vegetation can also still obstruct the camera’s view of the sand dune, with or without the use of HDR images. Pole aerial photography, in which the camera is mounted on an extended painter’s pole, can help avoid vegetation and provide higher-angle views of dune surfaces. HDR images, however, cannot be used with a hand-held pole due to movement of the camera once it is mounted on top of the pole. Processing HDR images is also time consuming, but the ending model is overall a better representation of the sand dune surface. HDR images can therefore aid in overcoming difficulties when using photogrammetry to study sand dunes or other homogenous geological surfaces.