.png){kind=icon}
.png){kind=icon}
Return
A comprehensive review on beyond correlation: Deep learning revolutionizes ghost imaging for target recognition and reconstruction through scattering media
.png){kind=icon}
PRE
AI
A
C
B
M
S
刘
DOI：10.1016/j.optlaseng.2026.109695.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
A potent method for non-invasive optical sensing and visualization in challenging environments, ghost imaging (GI) is well-known for its resistance to turbulence and scattering. This review article covers the transition from conventional physics-driven models to contemporary data-driven techniques, offering a thorough overview of the quick developments and paradigm shifts in GI through scattering media. First, we systematically divide the field into two categories: data-driven networks that use deep learning (DL) for image retrieval and classification, and physics-driven methods that include active wavefront correction and passive correlation imaging. The paper devotes considerable attention to the emerging field of physics-enhanced networks, which combine the representational capability of DL with the interpretability of physical models in a synergistic way. We also critically analyze the significant open challenges that cover the whole field, such as basic physical constraints and realworld hardware integration concerns pertaining to system stability and real-time performance. We conclude by outlining future directions, stressing the importance of standardized datasets, applications-driven benchmarking, and a general paradigm shift toward task-oriented imaging that goes beyond straightforward image reconstruction. This review attempts to be a fundamental resource for scholars navigating the changing field of computational imaging through scattering media by combining the interactions of physics, algorithms, and hardware.
Keywords:
Gi through scattering
Scattering media
Enhanced reconstruction
Deep learning
Traditional models
Journal
IF:
3.7
Papers:
7.1K
Citations:
1.7W
Organization
Cited Papers