A Review Toward Deep Learning for High Dynamic Range Reconstruction DOI Creative Commons
Gabriel de Lima Martins, Josue Lopez-Cabrejos,

J. P. S. Martins

et al.

Applied Sciences, Journal Year: 2025, Volume and Issue: 15(10), P. 5339 - 5339

Published: May 10, 2025

High Dynamic Range (HDR) image reconstruction has gained prominence in a wide range of fields; not only is it implemented computer vision, but industries such as entertainment and medicine also benefit considerably from this technology due to its ability capture reproduce scenes with greater variety luminosities, extending conventional levels perception. This article presents review the state art HDR methods based on deep learning, ranging classical approaches that are still expressive relevant more recent proposals involving advent new architectures. The fundamental role high-quality datasets specific metrics evaluating performance algorithms discussed, well emphasizing challenges inherent capturing multiple exposures dealing artifacts. Finally, emerging trends promising directions for overcoming current limitations expanding potential real-world scenarios highlighted.

Language: Английский

A Review Toward Deep Learning for High Dynamic Range Reconstruction DOI Creative Commons
Gabriel de Lima Martins, Josue Lopez-Cabrejos,

J. P. S. Martins

et al.

Applied Sciences, Journal Year: 2025, Volume and Issue: 15(10), P. 5339 - 5339

Published: May 10, 2025

High Dynamic Range (HDR) image reconstruction has gained prominence in a wide range of fields; not only is it implemented computer vision, but industries such as entertainment and medicine also benefit considerably from this technology due to its ability capture reproduce scenes with greater variety luminosities, extending conventional levels perception. This article presents review the state art HDR methods based on deep learning, ranging classical approaches that are still expressive relevant more recent proposals involving advent new architectures. The fundamental role high-quality datasets specific metrics evaluating performance algorithms discussed, well emphasizing challenges inherent capturing multiple exposures dealing artifacts. Finally, emerging trends promising directions for overcoming current limitations expanding potential real-world scenarios highlighted.

Language: Английский

Citations

0