Computational ghost imaging of atmospheric turbulence based on a single variable convolution kernel

As an indirect imaging modality, computational ghost imaging (CGI) has good anti-turbulence properties, which has been extensively used in the field of atmospheric turbulence imaging. However, constraints of sampling rate hamper its further application. To address this issue, the model-driven deep learning (MDL) approach is proposed, and it shows excellent performance in the task of target reconstruction under different conditions. Nevertheless, constrained by the depth of the network structure, it requires a greater time cost for achieving the target high-quality reconfiguration. In this paper, we design a single-variable convolution kernel (SVCK) MDL approach and employ it in the field of atmospheric turbulence computational ghost imaging. Compared to traditional CGI and MDL methods, our approach significantly reduces the time required for reconstruction while guaranteeing the imaging quality. Simulation and experimental results show that our method is superior to the traditional CGI method for target reconstruction, and the imaging quality is comparable to the MDL method with greatly reduced reconstruction time. Thus, our results provide an effective way for high-quality imaging in atmospheric turbulence.