Extendible ghost imaging with high reconstruction quality in strong scattering medium

Imaging objects hidden inside the strongly scattering media based on bidirectional ghost imaging

We demonstrate a novel, to the best of our knowledge, method for imaging objects hidden inside the strongly scattering media based on bidirectional ghost imaging (GI). In this method, GI is performed separately on both sides of the object, resulting in two GI results. Through an autocorrelation operation to the two GI results, the convolution between the autocorrelation of the object and the point spread function (PSF) of the strongly scattering media can be recovered. Therefore, the object can be recovered by obtaining the PSF of the strongly scattering media through noninvasive measurement or numerical calculation. Simulation and experimental results show that bidirectional ghost imaging (BGI) can reconstruct high-quality images, particularly when the thickness of the strongly scattering media greatly exceeds the scattering mean free path.

Learning-based correction with Gaussian constraints for ghost imaging through dynamic scattering media

In this Letter, we propose a learning-based correction method to realize ghost imaging (GI) through dynamic scattering media using deep neural networks with Gaussian constraints. The proposed method learns the wave-scattering mechanism in dynamic scattering environments and rectifies physically existing dynamic scaling factors in the optical channel. The corrected realizations obey a Gaussian distribution and can be used to recover high-quality ghost images. Experimental results demonstrate effectiveness and robustness of the proposed learning-based correction method when imaging through dynamic scattering media is conducted. In addition, only the half number of realizations is needed in dynamic scattering environments, compared with that used in the temporally corrected GI method. The proposed scheme provides a novel, to the best of our knowledge, insight into GI and could be a promising and powerful tool for optical imaging through dynamic scattering media.

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.