Adaptive space carving with texture mapping

Abstract

Space carving reconstructs a 3D object from multiple images, but existing algorithms rely on a regular grid which makes poor use of memory. By using the image information, adaptive space carving uses a recursively generated structure which reduces memory requirements and thus allows a finer grid. After reconstruction, models are triangulated to facilitate texture mapping. Experimental results show the enhanced appearance of models reconstructed in this way. © Springer-Verlag Berlin Heidelberg 2005.