Photogrammetry and dense stereo matching approach applied to the documentation of The cultural heritage site of Kilwa (Saudi Arabia)
Year: 2009
Type: conference-paper
Abstract: Since a couple of years, several commercial solutions of dense stereo matching have been developed. This process offers a really cheap, flexible and accurate solution to get 3D point clouds and textured models. The calibration of the camera allows a subpixellar correlation for correctly textured objects. In order to define the limits of such a process for cultural heritage applications, the Photomodeler Scanner© software has been tested on an archaeological site. The French project of archaeology and epigraphy in the region of Kilwa (Saudi Arabia) is aimed on two main works: prospecting and excavating. During the annual mission of one month a lot of information is collected, but the analysis of this data could not be done on the go. During prospecting in the area of 25km radius, epigraphs and carvings have been registered in a GIS with links to simple pictures in the database. But to describe, analyze and interpret those epigraphs, simple pictures are often not sufficient. Laser scanning cannot be used in this case for several reasons: difficulty of access and conditions (such as desert wind, or heat), location of epigraphs or cost. Therefore, solutions based on Dense Stereo Matching have been investigated. Two kinds of models are presented in this paper. The first application is the modeling of epigraphs from a single stereopair during prospection. Another application has been established on the site of Kilwa, where a building has been photographed and modeled. For this model, we got about 50 pictures. We inserted it on a Digital Elevation Model of the site obtained by tacheometry in order to meet the needs of the architects and archaeologists and to make assumptions about the original condition of the site. The main advantage of this photogrammetric methodology is to get at the same time a point cloud (resolution depends on the size of the pixel on the object), and therefore a mesh, and oriented high resolution images. After processing, we can use the data exactly as a laser scanning point cloud, with really better raster information for textures.
Source: HAL
Institution: Sciences de l'Antiquité et du Moyen-Age
Language: English
Subfield: Geology
Field: Earth and Planetary Sciences
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W3040922419