Simultaneous estimation of 3D shape and 2D silhouettes from multi-viewpoint images

本研究提出了一种从多视点图像中同时估计三维形状和二维轮廓的新算法。关键的贡献是3D形状和2D轮廓之间的一致性约束，即3D形状的投影。与传统方法不同，该算法在三维形状重建方案中明确地模拟了这种投影约束。

This research proposed a new algorithm that estimates both 3D shape and 2D silhouettes from multi-viewpoint images. The key contribution is the consistency constraint between 3D shape and 2D silhouettes, i.e., the projections of the 3D shape. Unlike conventional methods, the proposed algorithm explicitly models this projective constraint in the 3D shape reconstruction scheme.