Non-rigid structure from motion with incremental shape prior

Non-rigid Structure from Motion with Diffusion Maps Prior

“…KSFM: The kernel non-rigid structure from motion [12]. IPCA: The incremental principal components analysis based method [17]. DM: The proposed method.…”

“…Initial shapes S and camera motion R are estimated by running a few iteration of the optimisation process in batch NRSfM using linear basis shapes model [17]. For each initial shape, Nyström extension is used for embedding these new samples into the reduced space.…”

Real-time 3D reconstruction of non-rigid shapes with a single moving camera

“…Although this is a promising direction, these methods remain slow [40,47] or assume rigid priors that are not always available [4]. Furthermore, these methods use the 2D tracking data as input, i.e., they do not normally compute the tracking and data association on-the-fly [40,47]. While non-rigid shapes have been previously registered in real time [39,42], these methods do not estimate the camera pose, assumed to be known in advance.…”

“…This is because they can only process the image sequence in a batch manner, after the complete acquisition, limiting their applicability in real-time and on-line applications. Only recently a few sequential approaches have been proposed [4,40,47]. Although this is a promising direction, these methods remain slow [40,47] or assume rigid priors that are not always available [4].…”

“…Only recently a few sequential approaches have been proposed [4,40,47]. Although this is a promising direction, these methods remain slow [40,47] or assume rigid priors that are not always available [4]. Furthermore, these methods use the 2D tracking data as input, i.e., they do not normally compute the tracking and data association on-the-fly [40,47].…”