Smith et al – Structure from Motion Photogrammetry in Physical Geography

In this article, Smith et al discuss a new advancement in the field of topographic data collection: the combined forces of Structure from Motion (SfM) and Multi-View Stereo (MVS) analysis. This system works tangentially as a way to acquire data from still images and then constituting those images into a 3d model using point clouds taken from these original images. These would be georeferenced with points for better accuracy.

This proves to be an exciting change. While the need for topographic data remains an important factor for many organisations, governments and people, there is often a barrier that prevents attaining this data due to the high cost of obtaining the software packages, tools and skills necessary to do an effective topographic survey. SfM-MVS works fine on any sort of digital camera, with Smith musing about the potential use of smart phone-collected data in the use of participatory surveying.

While reading this article, I wondered if this quick and easy method to collect data would phase out the need for expertise and in the process lose an expert’s viewpoint of topographic data. Many of the other methods discussed in Smith’s article seemed to prioritize accuracy and scope over speed and cheapness, which in many cases may prove crucial. I can imagine companies, in cost-cutting intentions, switch from a system such as lidar or dGPS to SfM-MVS, and thus lose crucial aspects of its previous topographic surveys, which may then lead to negative externalities.

Smith discusses that there is a trend emerging in which SfM-MVS is being marketed in software packages and taking on a ‘black-box’ format. Smith spends a long time discussing the various steps taken in this method as a way to subvert that potentially damaging practice. The way in which specific technologies can be used most effectively needs to be considered.

While the inaccuracies inherent in SfM-MVS will make it difficult to find a place alongside more traditional methods of topographic surveying, it nonetheless holds up an important role in certain scenarios. While it is inaccurate, it is cheap, quick and easy to use. Smith brings up the example of SfM-MVS being used in post-flood events as a way to quickly survey the affected areas. In cases of search-and-rescue emergency settings, small details can be emitted without much being lost.

The digital ecology in which all these different methods for topological surveying take part of must be understood as a whole, and ideally, the strengths and weaknesses of each method must be properly communicated to the surveyor.



Comments are closed.