Researchers developed a new "ghost imaging" method that uses AI to reconstruct high-resolution images from very limited data. This technique overcomes the limitations of traditional ghost imaging, allowing for better image quality and faster acquisition times, even in challenging conditions like low light or far-field imaging.

None identified

This paper presents a novel ghost imaging technique that achieves super-resolution by incorporating a deep neural network constraint. The proposed method demonstrates significant improvements in both spatial resolution and sampling ratio compared to traditional ghost imaging techniques. The experimental results are convincing and show potential for practical applications. However, some limitations such as the computational cost and the complex setup prevent a perfect score.