Fast View Interpolation from Stereo: Simpler can be Better
Fast View Interpolation from Stereo: Simpler can be Better
复制标题
立体快速视图插值：越简单越好
DOI：
--
复制
发表时间：
2008
期刊：
影响因子：
--
通讯作者：
S. Roy
中科院分区：
文献类型：
--
作者：
N.Martin;S. Roy
在本文中，我们建议仅依靠图像来生成新颖的视图，并回想一下为什么在简单场景的视图插值的背景下，完整场景的建模通常过于昂贵。我们研究通过视差图的前向和后向映射来实现视图插值的方法。我们提出了每种情况都需要更少的计算并给出更好的结果。与我们的预期相反，尽管提供了非常糟糕的视差图，但非常简单的立体算法可以产生非常令人信服的插值。我们建议用深度不连续性的概率模型来解释这一点。我们在为拟合真实图像统计数据而创建的合成数据上测试该模型，并与立体图像中广泛使用的图像进行比较[13]。在实践中，前向和后向映射方法可以依靠实时运行的简单立体算法来产生非常好的结果。获取一系列真实图像，以便准确比较插值图像，并使用标准指标来评估质量。
In this paper, we propose to rely only on images to generate novel views, and recall why modeling of the complete scene is often too expensive in the context of view interpolation of simple scenes. We investigate ways to achieve view interpolation by mean of forward and backward mapping of disparity maps. We present situations in which each one requires less computations and gives better results. Contrary to what we might expect, very simple stereo algorithms can produce very convincing interpolation despite providing really bad disparity maps. We propose to explain this with a probabilistic model of depth discontinuities. We test this model on synthetic data created to fit real image statistics and compare with images widely used in stereo[13]. In practice, forward and backward mapping methods can rely on simple stereo algorithms running in real time, to produce very good results. A sequence of real images was acquired to allow accurate comparison of interpolated images, and standard metrics are used to assess the quality.