STRUCTURE-FROM-MOTION PHOTOGRAMMETRY TECHNIQUES AND DIGITAL SURFACE MODEL CREATION FOR GEOSCIENCE APPLICATIONS

Structure-from-Motion Multi-View Stereo Photogrammetry (SfM or SfM-MVS) is a technique which produces high-resolution point cloud datasets from multiple overlapping photographs and a network of ground control points. In recent years, SfM has emerged as a viable remote sensing methodology for collecting lower-cost, high-resolution, three-dimensional geospatial data.

SfM has the potential to be a powerful option for the collection of three-dimensional remote sensing data, alongside traditional stereo photogrammetry, terrestrial lidar scanning, and airborne lidar scanning. In comparison to other survey methods, the primary benefits of SfM are significantly lower collection/production costs, portability of collection equipment when used in remote areas, and the flexibility to capture data at a broad range of spatial scales.

High-resolution digital terrain models are a valuable resource for many geoscience applications, but are commonly available only where airborne lidar has been collected. Prior to SfM photogrammetry and in the absence of available lidar, the best option available for constructing digital terrain models has usually been 1/3 arc-second (~10m) resolution National Elevation Data (NED), which presents a significant gap in resolution. Currently, much of eastern Oregon lacks lidar surveys, presenting a challenge for obtaining digital surface models necessary for geoscience projects. SfM derived digital surface models fill the need for accurate high-resolution data sets, replacing the currently available low resolution digital elevation models in areas lacking airborne lidar surveys.

Oregon Department of Geology and Mineral Industries staff have developed a successful SfM workflow that utilizes National Agricultural Imagery Program (NAIP) aerial photos to produce sub-meter resolution digital surface models for areas of interest in sparsely vegetated parts of eastern Oregon. Multiple-year, repeat acquisitions of NAIP stereo-pair aerial photos provide complete coverage for the State of Oregon. The use of NAIP imagery introduces a number of manageable challenges to SfM processing, but ultimately allows for the creation of high-resolution digital terrain models of consistent quality for areas lacking such coverage.