Ptychography-enhanced correlography for non-line-of-sight imaging

Non-line-of-sight (NLOS) imaging offers promising solutions for extending the observable field of view in applications such as autonomous driving, remote sensing, and medical diagnostics. However, existing steady-state NLOS imaging methods struggle to reconstruct complex hidden objects due to incomplete sampling of the object’s Fourier amplitude. To address this challenge, we propose a ptychography-enhanced correlography for NLOS imaging (PEC-NLOS). This method leverages indirect speckle illumination on the relay wall and uses an industrial polarization-sensitive camera to acquire partial Fourier amplitude measurements of the hidden object. By scanning multiple indirect illumination points, a series of overlapping speckle patterns is collected. Through iterative phase retrieval using a ptychographic reconstruction algorithm, PEC-NLOS enables high-resolution and high-fidelity reconstruction of complex hidden scenes. Moreover, we introduce a 3D data correction strategy based on optical modeling of the NLOS geometry, incorporating non-invasive positional priors to refine all acquisition data. We not only address the limitations of complex object reconstruction and ambiguity in steady-state NLOS imaging but also establish a foundation for robust and efficient NLOS imaging in complex real-world environments.