Terrain measurement using automated digital photogrammetry

Abstract

IntroductionPhotogrammetry has traditionally provided a means of generating three-dimensional spatial data to represent terrain surfaces, which complements traditional ground-based surveying methods. Although techniques such as airborne laser scanning (Lohr 1998) and synthetic aperture radar (Hogg et al. 1993; Vencatasawamy et al. 1998) have developed, photogrammetry remains the primary method of generating topographic maps (Wolf 1983; Capes 1998). One important advantage of photogrammetry is the flexibility of scale that allows application to imagery acquired from ground, air and space. Indeed, a new generation of high (i.e. 1 m) resolution satellite sensors (Capes 1998) is likely to further increase the potential applications of photogrammetry. Despite many advantages, there have been several problems with the application of photogrammetry using traditional methods. Most significantly, there was the requirement to use an expensive and complex photogrammetric stereoplotter. This ensured that the measurement process was slow and generally required the skills of an experienced operator, particularly if results of the highest accuracy were to be obtained.Rapid developments in computing hardware and software have allowed the science of photogrammetry to develop rapidly during the last ten years (Gruen 1994; Atkinson 1996; Greve 1996). These developments have radically eased many of the problems and limitations associated with traditional analogue instrumentation. Use of a purely numerical or analytical solution provides flexibility, which assists in two important ways. Satellite imagery, oblique aerial photography and ground-based imagery can be used, in addition to the more traditional vertical aerial perspective. Similarly, imagery acquired using