Abstract

Recovering the 3D shape of deformable surfaces from sin- gle images is difficult because many different shapes have very similar pro jections. This is commonly addressed by restricting the set of possi- ble shapes to linear combinations of deformation modes and by imposing additional geometric constraints. Unfortunately, because image measure- ments are noisy, such constraints do not always guarantee that the correct shape will be recovered. To overcome this limitation, we introduce an ef- ficient approach to exploring the set of solutions of an ob jective function based on point-correspondences and to proposing a small set of candidate 3D shapes. This allows the use of additional image information to choose the best one. As a proof of concept, we use either motion or shading cues to this end and show that we can handle a complex ob jective function without having to solve a difficult non-linear minimization problem.