Visual attention-based 3D multiple LOD modeling for virtual environments

The main objective of this paper is to contribute to the advancement of 3D modeling techniques in order to make them more adapted to human visual capabilities. The proposed approach exploits human visual attention characteristics and incorporates them in a multiple level-of-detail (LOD) 3D object modeling scheme in order to ensure a better quality of simplified models at various resolutions. A visual attention computational model is initially employed to detect points of interest over the surface of a 3D object. An adapted version of a classical simplification algorithm performs simplifications at different resolutions while preserving the regions identified as important by the visual attention model. The quality of the resulting models is compared with the performance of other interest point detectors incorporated in a similar manner in the simplification algorithm. The proposed solution performs better, especially at lower resolution, and results overall in higher quality models.