Reconstruction method of computationally convolutional ghost imaging based on total variation regularization

To address the problems of severe noise and blurred edge structures in the reconstruction results of computationally convolutional ghost imaging (CCGI) under low sampling rates, this paper introduces total variation regularization and incorporates it into the CCGI reconstruction framework, thereby proposing a TV-based CCGI reconstruction method (TV-CCGI). By establishing a unified imaging model and combining numerical simulations with experimental validation, a systematic comparative analysis of the reconstruction performance of CCGI and TV-CCGI is carried out. The experimental results demonstrate that, under the same sampling rate, the proposed TV-CCGI method achieves significant improvements in noise suppression, preservation of target edge structures, and overall reconstruction quality. In particular, these advantages become more pronounced under low sampling rates, which verifies the effectiveness, stability, and superiority of the proposed method.