Unsupervised disparity estimation from light field using plug-and-play weighted warping loss
作者：
Highlights：
• We address the problem of unsupervised disparity estimation from a light field.
• We construct our method via a plug-and-play approach by replacing the loss function.
• Our loss function is designed to be aware of occlusions and edge alignment.
• Our method achieves state-of-the-art performance as an unsupervised method.
• Our method can obtain high-quality disparity maps from natural scenes.
摘要
•We address the problem of unsupervised disparity estimation from a light field.•We construct our method via a plug-and-play approach by replacing the loss function.•Our loss function is designed to be aware of occlusions and edge alignment.•Our method achieves state-of-the-art performance as an unsupervised method.•Our method can obtain high-quality disparity maps from natural scenes.
论文关键词：Light field,Disparity estimation,CNN,Unsupervised learning
论文评审过程：Received 25 August 2021, Revised 9 May 2022, Accepted 1 June 2022, Available online 8 June 2022, Version of Record 17 June 2022.
论文官网地址：https://doi.org/10.1016/j.image.2022.116764
