Hologram super-resolution network for improved fringe quality toward higher phase reconstruction accuracy in digital holographic microscopy.

Digital holographic microscopy (DHM) has shown great potential in biomedical imaging and quantitative phase analysis due to its wide field of view, non-contact operation, and high measurement sensitivity. However, in practical DHM systems, the quality of the recorded holograms is not only limited by the optical imaging conditions but also constrained by the finite pixel pitch and pixel integration effect of the image sensor. These factors may reduce the spatial resolution of holographic fringe images, making it difficult to capture high-frequency information from the sample and thereby affecting the accuracy of phase reconstruction. To address this issue, we propose a multi-stage perceptual super-resolution network with directional variance attention for hologram enhancement under digital sampling constraints. The proposed method aims to improve the quality of holographic fringe patterns from low-resolution holograms, thereby enhancing the accuracy of phase reconstruction. Experimental results demonstrate that the proposed algorithm can effectively improve the quality of holographic fringe patterns in holographic imaging systems without requiring any advanced physical hardware and achieve more accurate phase reconstruction than competing methods. These results indicate that the proposed framework serves as an effective computational enhancement tool for improving phase reconstruction quality in DHM.