Color-faithful differentiable Fourier ptychographic microscopy

Fourier ptychographic microscopy (FPM) enhances the spatial-bandwidth product of imaging systems, but its extension to color imaging remains challenging. Existing channel-split methods incur long acquisition times, while multiplexed-illumination approaches suffer from color leakage and require complex post-processing. In this work, we introduce a novel, to the best of our knowledge, color-faithful FPM framework under multiplexed illumination by integrating a physically interpretable color constraint into a differentiable FPM model. Unlike prior works, our method enforces color fidelity during optimization by constraining the first- and second-order moment of RGB channels, effectively preserving the true color tone in reconstructions. This statistical constraint not only improves color accuracy but also offers broad compatibility with existing differentiable FPM models. Numerical simulations and optical experiments show that our method ensures high color fidelity, providing a practical full-color FPM solution.