3D Information from Scattering Media Images

Research output: Contribution to journalArticlepeer-review

Abstract

Scattering media environments are real-world conditions that occur often, in daily life. Some examples of scattering media are haze, fog, and other bad weather conditions. In these environments, micro-particles in the surrounding media interfere with light propagation and image formation. Thus, images that are captured in these scattering media environments will suffer from low contrast and loss of intensity. This becomes an issue for computer vision methods that employ features found in the scene. To solve this issue, many approaches must estimate the corresponding clear scene prior to further processing. However, the image formation model in scattering media shows potential 3D distance information about the scene encoded implicitly in image intensities. In this paper, we investigate the potential information that can be extracted directly from the scattering media images. We demonstrate the possibility of extracting relative depth in the form of transmission as well as explicit depth maps from single images.
Original languageEnglish
Pages (from-to)73-82
JournalJurnal Ilmu Komputer dan Informasi
Volume14
Issue number1
DOIs
Publication statusPublished - 28 Feb 2021

Keywords

  • scattering media
  • 3D depth
  • transmission
  • statistical prior
  • airlight

Fingerprint

Dive into the research topics of '3D Information from Scattering Media Images'. Together they form a unique fingerprint.

Cite this