Image-based rendering from a sparse set of images
Image-based rendering from a sparse set of images
复制标题
从一组稀疏图像进行基于图像的渲染
DOI：
10.1145/1187112.1187290
复制
发表时间：
2005
期刊：
影响因子：
--
通讯作者：
P. Belhumeur
中科院分区：
文献类型：
--
作者：
Todd E. Zickler;S. Enrique;R. Ramamoorthi;P. Belhumeur
当物体的形状已知时，其外观由定义在其表面上的空间变化的双向反射分布函数（SBRDF）确定。我们提出了一种方法，用于从稀疏图像集恢复已知几何形状的表面的SBRDF。与现有的参数方法（例如，[Sato等人，1997]）这种方法不需要选择分析BRDF模型，因此能够捕获任意复杂的反射率效应。它也不同于现有的非参数方法（例如[Matusik et al. 2002]），这些方法纯粹是数据驱动的，可能需要数千张图像;与这些方法不同的是，我们利用空间相干性，这可以大大减少所需输入图像的数量。
When the shape of an object is known, its appearance is determined by the spatially-varying bi-directional reflectance distribution function (SBRDF) defined on its surface. We present a method for recovering the SBRDF of a surface with known geometry from a sparse set of images. Unlike existing parametric methods (e.g., [Sato et al. 1997]) this approach does not require a choice of an analytic BRDF model, and is therefore capable of capturing arbitrary, complex reflectance effects. It is also different from existing non-parametric methods (e.g. [Matusik et al. 2002]) that are purely data-driven and can require thousands of images; unlike these methods, we exploit spatial coherence, which can drastically reduce the number of required input images.