Noise-resilient single-pixel compressive sensing with single photon counting

This paper investigates the degradation of single-pixel optical classifiers using compressive sensing due to photon-counting noise and proposes using quantum parametric mode sorting (QPMS) to restore performance. Using MNIST handwritten digits, the effects of detector dark counts and in-band background noise are examined, demonstrating the effectiveness of mode filtering and upconversion detection. The study achieves 94% classification accuracy even with 500 times stronger in-band noise than the received signal, suggesting a robust approach for single-photon sensing in noisy environments.

Lili Li, Santosh Kumar, Yong Meng Sua, Yu-Ping Huang