Hardware Accelerated Voxel Carving
Miguel Ángel Sainz, Nader Bagherzadeh, Antonio Susín · 2002
In this paper we present a fast hardware accelerated volumetric method for the reconstruction of real world objects from a set of calibrated images. The approach is based on carving a bounding volume using a color similarity criterion. The algorithm is designed to use hardware accelerated features from the videocard. Moreover, the data structures have been highly optimized in order to minimize run-time memory usage. Additional techniques such as hardware texture mapping and shadow polygons are used to avoid redundant calculations. Hardware texture mapping is also used to reach subpixel precision. Moreover, if the image data can be segmented into foreground and background, a real time implementation of the visual hull algorithm is added to achieve a substantial reduction in computation time. Experimental results are presented on both synthetic and real imagery, showing excellent visual quality of the volumetric reconstruction.