International Journal of Computer Vision manuscript No. (will be inserted by the editor) A Bayesian Approach to Multi-view 4D Modeling

Abstract This paper considers the problem of auto-matically recovering temporally consistent animated 3D models of arbitrary shapes in multi-camera setups. An approach is presented that takes as input a sequence of frame-wise reconstructed surfaces and iteratively de-forms a reference surface such that it fits the input observations. This approach addresses several issues in this field that include: large frame-to-frame deforma-tions, noise, missing data, outliers and shapes com-posed of multiple components with arbitrary geome-tries. The problem is cast as a geometric registration with two major features. First, surface deformations are modeled using mesh decomposition into elements called patches. This strategy ensures robustness by enabling flexible regularization priors through inter-patch rigid-ity constraints. Second, registration is formulated as a Bayesian estimation that alternates between probabilis-tic datal-model association and deformation parameter estimation. This accounts for uncertainties in the ac-quisition process and allows for noise, outliers and miss-ing geometries in the observed meshes. In the case of marker-less 3D human motion capture, this framewor