Model Evolution: An Incremental Approach to Non-Rigid Structure from Motion

Shengqi Zhu Li Zhang Brandon M. Smith

University of Wisconsin, Madison



We present a new framework for non-rigid structure from motion (NRSFM) that, for the first time in the literature, simultaneously addresses three significant challenges: severe occlusion, perspective camera projection, and large non-linear deformation. We introduce a concept called a model graph, which greatly reduces the computational cost of discovering groups of input images that depict consistent 3D shapes. A 3D model is constructed for each input image by traversing the model graph along multiple evolutionary paths. A compressive shape representation is constructed, which (1) consolidates the multiple 3D models for each image reconstructed during model evolution and (2) reduces the number of models needed to represent the input image set. Assuming feature correspondences are known, we demonstrate our algorithm on both real and synthetic data sets that exemplify all three aforementioned challenges.

Shengqi Zhu, Li Zhang, Brandon M. Smith, Model Evolution: An Incremental Appraoch to Non-Rigid Structure from Motion, IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR), June, 2010. [PDF 1.1MB]
This work is supported in part by National Science Foundation IIS-0845916 and IIS-0916441, and a Sloan Research Fellowship. Brandon Smith is also supported by a NSF graduate fellowship.
Supplemental Video
Download [MP4 32.9 MB]
Motion Evolution Video


Extended Video With a New Experiment
Download [MP4 74.5 MB]
Extended Motion Evolution Video


Code and More Examples are coming...