Stitching and Processing Gnomonic Projections for Close-Range Photogrammetry

This paper presents a 3D reconstruction methodology based on gnomonic projections generated from multiple central perspectives. The method can be useful when traditional images are insufficient to capture fine portions of objects with an adequate level of detail, especially in close-range photogrammetry. The aim of this paper is to prove that gnomonic projections are powerful tools to recover small details that cannot be reconstructed from standard images. The generation of the gnomonic projections is discussed, as well as the methodology to compute camera parameters, orient multiple projections, and finally extract textured 3D models or orthophotos. To verify the correctness of the methodology, some comparisons with sets of independent checkpoints were carried out. The accuracy assessment confirmed the correctness of the mathematical approach and underlined how gnomonic projections are a valid alternative to standard images. The method itself has been already used in surveys for cultural heritage documentation.