Fast Light-Weight Near-Field Photometric Stereo
Fast Light-Weight Near-Field Photometric Stereo
复制标题
DOI：
10.1109/cvpr52688.2022.01228
复制
发表时间：
2022-03
期刊：
影响因子：
--
通讯作者：
Daniel Lichy;Soumyadip Sengupta;D. Jacobs
中科院分区：
文献类型：
--
作者：
Daniel Lichy;Soumyadip Sengupta;D. Jacobs
我们将第一个基于学习的端到端解决方案引入近场光度立体（PS），其中光源靠近感兴趣的对象。此设置对于重建大型固定对象特别有用。我们的方法速度很快，在商品GPU上大约1秒内从52个512 x384分辨率的图像中生成网格，因此可能会解锁几个AR/VR应用程序。现有的方法依赖于与在像素或小块上操作的远场PS网络相结合的优化。使用优化使这些方法缓慢且内存密集（需要17 GB GPU和27 GB CPU内存），而仅使用像素或补丁使它们极易受到噪声和校准错误的影响。为了解决这些问题，我们开发了一个递归的多分辨率方案，估计表面法线和深度图的整个图像在每一步。然后使用每个尺度下的预测深度图来估计下一尺度的每像素照明。这种设计使我们的方法比使用迭代优化的最先进的近场PS重建技术快近45倍，精度高2°（11.3°与13.3°平均角误差）。
We introduce the first end-to-end learning-based solution to near-field Photometric Stereo (PS), where the light sources are close to the object of interest. This setup is especially useful for reconstructing large immobile objects. Our method is fast, producing a mesh from 52 512x384 resolution images in about 1 second on a commodity GPU, thus potentially unlocking several AR/VR applications. Existing approaches rely on optimization coupled with a far-field PS network operating on pixels or small patches. Using optimization makes these approaches slow and memory intensive (requiring 17GB GPU and 27GB of CPU memory) while using only pixels or patches makes them highly sus-ceptible to noise and calibration errors. To address these issues, we develop a recursive multi-resolution scheme to estimate surface normal and depth maps of the whole image at each step. The predicted depth map at each scale is then used to estimate 'per-pixel lighting, for the next scale. This design makes our approach almost 45x faster and 2° more accurate (11.3° vs. 13.3° Mean Angular Error) than the state-of-the-art near-field PS reconstruction technique, which uses iterative optimization.