Advances in ghost imaging based on deep learning

Ghost imaging is a rapidly developing imaging technique that has garnered increasing attention in recent years due to its strong anti-interference capability, ability to operate under low-light conditions, and relatively simple system configuration. However, its practical implementation is still constrained by several technical challenges, including low signal-to-noise ratio (SNR), slow image reconstruction speed, and stringent requirements for system stability. In recent years, the rapid advancement of deep learning (DL) has offered new avenues for addressing these limitations. Leveraging the nonlinear fitting capacity and automatic feature extraction capability of neural networks, researchers have made significant progress in image reconstruction, target segmentation, and object recognition. This review systematically outlines the theoretical foundations of ghost imaging, critically examines the current applications and key achievements of DL in this field, and analyzes the existing challenges. Furthermore, it explores future directions for the deep integration of DL with ghost imaging technologies. This work aims to serve as a comprehensive reference for researchers and to promote the intelligent and practical development of ghost imaging.