Multiframe stacked integration for high-fidelity single-pixel imaging via weighted fractional-order Hermite-Gaussian modulation

Conventional modulation schemes in single-pixel imaging (SPI) are hindered by suboptimal encoding efficiency and inherent optical mismatches, ultimately impeding imaging resolution and fidelity. This work presents an advanced modulation framework leveraging fractional-order Hermite-Gaussian (FrHG) bases for precise light-field encoding. Imaging reconstruction is achieved via dual-difference correlations between photon fluctuations and modulated fields. To counteract quantization artifacts, pixel-weighted preprocessing informed by error priors ensures accurate spatial correspondence. By enabling precise registration of multiframe fractional reconstructions through partitioned stacking scan, a Bayesian-based feature fusion strategy further refines global detail, robustly delivering artifact-free spatial contrast. Experiments across diverse microscopic scenarios including textured tissue, demonstrate the superior performance of proposed method. This approach is compatible with various SPI hardware architectures, holding promise as an adaptable, high-resolution solution applicable to non-visible and low-light microscopy.