Low-cost model reconstruction from image sequences

摘要

A system that constructs a three dimensional model using two dimensional images taken from multiple view-points is presented. This system improves upon existing work by including several optimisations and extensions to cater for poor lighting. This system was developed with the modeling of African artworks in mind. As these artifacts are often located in remote areas, our system has to be robust enough to deal with less than ideal lighting conditions.The images used as input are obtained by filming an object, which is being rotated on a turntable. The modeling process begins with the extraction of "silhouettes" from the input images. These silhouettes are used in conjunction with the camera model to construct a volumetric representation. Following this, a surface model, consisting of a triangular mesh, is created. Finally, the surface model is texture mapped.