arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2602.22279eess.IVcs.AIcs.SD
收藏

从饱和数据中学习重建:音频去削波与高动态范围成像

Learning to reconstruct from saturated data: audio declipping and high-dynamic range imaging

CNRS and ENS de Lyon, Laboratoire de physique(CNRS和里昂大学,物理实验室);UCLouvain, ICTEAM(乌斯特大学,ICTEAM)

Victor Sechaud, Laurent Jacques, Patrice Abry, Julián Tachella

AI总结:

本文提出了一种基于自监督学习的方法,用于从削波测量中恢复音频和图像,通过假设信号分布对幅度变化不变,实现了与全监督方法相当的重建效果。

摘要:

基于学习的方法现在普遍用于解决逆问题,但在实际应用中,其部署常受到缺乏地面真实参考的限制。最近的自监督学习策略提供了一个有前途的替代方案,避免了对地面真实的需要。然而,大多数现有方法局限于线性逆问题。本工作将自监督学习扩展到非线性问题,即从削波测量中恢复音频和图像,通过假设信号分布近似对幅度变化不变。我们为仅从饱和信号中学习重建提供了充分条件,并提供了一种可用于训练重建网络的自监督损失。在音频和图像数据上的实验表明,所提出的方法在仅依赖削波测量进行训练的情况下,效果几乎与全监督方法相当。

英文摘要:

Learning based methods are now ubiquitous for solving inverse problems, but their deployment in real-world applications is often hindered by the lack of ground truth references for training. Recent self-supervised learning strategies offer a promising alternative, avoiding the need for ground truth. However, most existing methods are limited to linear inverse problems. This work extends self-supervised learning to the non-linear problem of recovering audio and images from clipped measurements, by assuming that the signal distribution is approximately invariant to changes in amplitude. We provide sufficient conditions for learning to reconstruct from saturated signals alone and a self-supervised loss that can be used to train reconstruction networks. Experiments on both audio and image data show that the proposed approach is almost as effective as fully supervised approaches, despite relying solely on clipped measurements for training.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑