In this paper, we present a new area-based stereo matching algorithm using adaptive binary window. While many area-based algorithms have been proposed in recent years, the selection and computation of the size and shape of the matching window is still a difficult problem and an obstacle in real-time implementation of these algorithms. We develop adaptive binary window and different window matching approach which gives very good results and reduces the algorithm's running time to milliseconds. Our experiments show that unlike other area-based algorithms, our method works very well at disparity boundaries as well as in low textured areas and computes a sharp disparity map. Evaluation on the benchmark Middlebury stereo dataset shows that the performance of our algorithm is the best as compared to other local as well as global real-time stereo matching algorithms.

Presented by Raj Kumar Gupta

In this paper we present a method to iteratively capture the dynamic evolution of a surface from a set of point clouds independently acquired from multi-view videos. Without prior knowledge on the observed shape we deform the first reconstructed mesh across the sequence to fit these point clouds while preserving the local rigidity with respect to this reference pose. The deformation of this mesh is guided by control points that are randomly seeded on the surface, and around which rigid motions are locally averaged. These rigid motions are computed by iteratively re-establishing point-to-point associations between the deformed mesh and the target data in a way inspired by ICP. Experimental results, including quantitative analysis, on standard and challenging datasets obtained from real video sequences show the robustness and the precision of the proposed scheme.

Presented by Cedric Cagniart