Abstract
A method for variational modeling of individual hands from a template hand model is described. The marked individual hand is reconstructed based on image-based multi-view 3D model reconstruction technique. According to hand joint anatomical knowledge, a single axis rotation assumption is proposed for estimating the joint axes and joint centers by the positions of dot markers at the backside of the hand skin in different hand postures. On the other hand, the skin surface of the template hand model is hierarchically partitioned into segments according to the hand surface anatomy by main creases on palmar side. The segments of the template hand model are scaled according to the feature dimensions of the reconstructed individual hand model. Finally, the template hand model scaled in segment-wise way is registered to the reconstructed hand model by a rough alignment process where the gravity center of the point sets on the creases of both models are matched and a fine alignment process where the segments of both models are matched based on ICP algorithm. Compared with previous works, the method is more subject-friendly, inexpensive, and efficient.
