自用学习,欢迎指正。 《Deep Photometric Stereo for Non-Lambertian Surfaces》 = 《PS-FCN:A Flexible Learning Framework for Photometric Stereo》+ 《Self-calibrating Deep Photometric Stereo Networks》 作者: Guanying Chen Th
本文主要研究了深度学习在非朗伯表面光度立体问题的应用。具体来说，论文提出了两种方法，一种是深度光度立体框架（PS-FCN），用于已标定的表面，另一种是LC-NET，用于非标定表面，旨在预测未知光照方向。PS-FCN是深度学习方法解决光度立体问题的创新尝试，它能直接学习物体表面观测到的辐射值与法向之间的映...
Photometric stereo for Lambertian surfaces For a scene consisting of a Lambertian surface illuminated by a single distant (parallel ray) light source, the observed image intensity I at each pixel can be simply written as the product of the composite albedo KD and the cosine of the incidence angl...
History 3 Commits quantify_reflections.py Initial commit Jul 30, 2024 Repository files navigation README SPINS Code forthe IEEE OJCS paper Non-Lambertian Surfaces and Their Challenges for Visual SLAM (2024). The code computes the quantity of non-Lambertian surfaces for the SPINS dataset. ...
3 Commits README.md Updated README.md Feb 12, 2025 quantify_reflections.py Initial commit Jul 30, 2024 quantify_reflections_hsv.py Initial commit Jul 30, 2024 Repository files navigation README SPINS Code forthe IEEE OJCS paper Non-Lambertian Surfaces and Their Challenges for Visual SLAM (2024...