摘要

Ortho-photo is one of the most important photogrammetric products for archaeological documentation. It consists of a powerful textured representation combining geometric accuracy with rich detail, such as areas of damage and decay. Archaeological applications are usually faced with complex object shapes. Compared with conventional algorithms, ortho-projection of such rough curved objects is still a problem, due to the complex description of the analytical shape of the object. Even using a detailed digital surface model, typical ortho-rectification software does not produce the desired outcome, being incapable of handling image visibility and model occlusions, since it is limited to 2.5-dimensional surface descriptions. This paper presents an approach for the automated production of true ortho-mosaics for the documentation of cultural objects. The algorithm uses precise three-dimensional surface representations derived from laser scanning and several digital images that entirely cover the object of interest. After identifying all model surface triangles in the viewing direction, the triangles are projected back on to all initial images to establish visibilities for every available image. Missing image information can be filled in from adjacent images that must have been subjected to the same true ortho-photo procedure.

  • 出版日期2010-6