We introduce a 3D reconstruction technique based on near-light photometric stereo, where lighting calibration is achieved on-the-fly from a rough geometric knowledge obtained, e.g., by structure-from-motion. From this coarse geometry, the proposed graduated optimisation setup estimates all light source parameters (position, intensity, orientation, and anisotropy factor of each source), which can then be used within a calibrated photometric stereo algorithm. A series of real-world experiments is conducted, which validates the interest of the proposed approach.

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

Self-calibrated Near-Light Photometric Stereo Using a Geometric Proxy

  • Benjamin Coupry,
  • François Lauze,
  • Antoine Laurent,
  • Jean Melou,
  • Yvain Quéau,
  • Jean-Denis Durou

摘要

We introduce a 3D reconstruction technique based on near-light photometric stereo, where lighting calibration is achieved on-the-fly from a rough geometric knowledge obtained, e.g., by structure-from-motion. From this coarse geometry, the proposed graduated optimisation setup estimates all light source parameters (position, intensity, orientation, and anisotropy factor of each source), which can then be used within a calibrated photometric stereo algorithm. A series of real-world experiments is conducted, which validates the interest of the proposed approach.