Key points are not available for this paper at this time.
Conservation of natural ecosystems requires regular monitoring of biodiversity, including the estimation of wildlife density. Recently, unmanned aerial systems (UAS) have become more available for numerous civilian applications. The use of small drones for wildlife surveys as a surrogate for manned aerial surveys is becoming increasingly attractive and has already been implemented with some success. This raises the question of how to process UAS imagery in order to determine the surface area of sampling strips within an acceptable confidence level. For the purpose of wildlife surveys, the estimation of sampling strip surface area needs to be both accurate and quick, and easy to implement. As GPS and an inertial measurement units are commonly integrated within unmanned aircraft platforms, two methods of direct georeferencing were compared here. On the one hand, we used the image footprint projection (IFP) method, which utilizes collinearity equations on each image individually. On the other hand, the Structure from Motion (SfM) technique was used for block orientation and georeferencing. These two methods were compared on eight sampling strips. An absolute orientation of the strip was determined by indirect georeferencing using ground control points. This absolute orientation was considered as the reference and was used for validating the other two methods. The IFP method was demonstrated to be the most accurate and the easiest to implement. It was also found to be less demanding in terms of image quality and overlap. However, even though a flat landscape is the type most widely encountered in wildlife surveys in Africa, we recommend estimating IFP sensitivity at an accentuation of the relief.
Lisein et al. (Sun,) studied this question.