Differentially sampling the transmission matrix for enhanced imaging in multimode fibers

Transmission matrices are essential for achieving high-fidelity wavefront shaping and correction in complex media. Despite this, current methods for sampling transmission matrices are susceptible to errors introduced by spurious interference, drift, and field reconstruction methods. In this work, we present a robust differential algorithm that is capable of overcoming these common challenges. Ultimately, our differential algorithm in the presence of common-mode noise and drift yields improvements of 40 to over 100% in field reconstruction fidelity for transmission through a multimodal fiber when compared to previous methods with uncompensated common-mode noise contributing 0.3% of the total input power. While results are presented for a multimodal fiber system, our methodology is applicable to any complex media, paving the way for robust wide-field imaging in real-world applications.