Semantic ghost imaging based on recurrent-neural-network

Fast focusing method in ghost imaging with a tracking trajectory

The imaging environment is unstable for trembling disturbance, which is detrimental to object reconstruction. In this Letter, we experimentally investigated ghost imaging (GI) under a temporal trembling disturbance. The fast-focusing method based on imaging with small sampling measurements is proposed, and the theoretical model and algorithm are validated. It is demonstrated that the proposed method is effective to obtain a better-resolution image of the object under the strong trembling disturbance including a laboratory and a real trembling environment. The results provide a promising approach to deal with image degradation caused by an unstable environment and can find potential applications for ghost imaging in remote sensing.

High-quality and high-diversity conditionally generative ghost imaging based on denoising diffusion probabilistic model

Deep-learning (DL) methods have gained significant attention in ghost imaging (GI) as promising approaches to attain high-quality reconstructions with limited sampling rates. However, existing DL-based GI methods primarily emphasize pixel-level loss and one-to-one mapping from bucket signals or low-quality GI images to high-quality images, tending to overlook the diversity in image reconstruction. Interpreting image reconstruction from the perspective of conditional probability, we propose the utilization of the denoising diffusion probabilistic model (DDPM) framework to address this challenge. Our designed method, known as DDPMGI, can not only achieve better quality but also generate reconstruction results with high diversity. At a sampling rate of 10%, our method achieves an average PSNR of 21.19 dB and an SSIM of 0.64, surpassing the performance of other comparison methods. The results of physical experiments further validate the effectiveness of our approach in real-world scenarios. Furthermore, we explore the potential application of our method in color GI reconstruction, where the average PSNR and SSIM reach 20.055 dB and 0.723, respectively. These results highlight the significant advancements and potential of our method in achieving high-quality image reconstructions in GI, including color image reconstruction.

Single-Pixel Imaging in Space and Time with Optically Modulated Free Electrons

Single-pixel imaging, originally developed in light optics, facilitates fast three-dimensional sample reconstruction as well as probing with light wavelengths undetectable by conventional multi-pixel detectors. However, the spatial resolution of optics-based single-pixel microscopy is limited by diffraction to hundreds of nanometers. Here, we propose an implementation of single-pixel imaging relying on attainable modifications of currently available ultrafast electron microscopes in which optically modulated electrons are used instead of photons to achieve subnanometer spatially and temporally resolved single-pixel imaging. We simulate electron beam profiles generated by interaction with the optical field produced by an externally programmable spatial light modulator and demonstrate the feasibility of the method by showing that the sample image and its temporal evolution can be reconstructed using realistic imperfect illumination patterns. Electron single-pixel imaging holds strong potential for application in low-dose probing of beam-sensitive biological and molecular samples, including rapid screening during in situ experiments.