The University of British Columbia

Effects of sampling rate and pose error on silhouette-based volume reconstruction are addressed in this paper. An account of importance of choosing silhouette samples (viewpoint distribution) is given. A new viewpoint distribution method is introduced and its performance is compared to a traditional method empirically. We also investigate effects of sampling rate variation and different levels of pose error on silhouette-based reconstruction quality. We observed that in presence of considerable pose error, increase of sampling rate cannot enhance reconstruction quality. We also observed that reconstruction process in higher sampling rates is more sensitive to pose error.