Single multimode fiber imaging based on low-rank recovery

Abstract We present a novel scheme of single multimode fiber imaging by exploiting low-rank constraint for a faithful image recovery. Compared with the commonly-adopted sparsity constraint, the proposed scheme takes the advantage of the self-similarity property of natural images, and is demonstrated to achieve higher image quality and smoothness in multimode fiber imaging, especially in the under-sampling cases. We also demonstrate the robustness of this method to recover different kind of images against fiber bending, and discuss the low-rank parameter tunning for a stable recovery. Our findings may pave the way for a simpler and more robust scheme of single multimode fiber imaging, which could be applied for many demanding imaging tasks.