Motion-induced phase-shifting profilometry for dynamic objects via Fourier fringe analysis and speckle correlation-assisted phase unwrapping.

Fringe projection profilometry, particularly phase-shifting profilometry, has been extensively studied and widely adopted due to its non-contact operation, high accuracy, and efficiency. However, in dynamic applications, it faces two major challenges: the increased number of encoded patterns required for absolute phase retrieval, and motion-induced phase errors. To address these limitations while preserving measurement accuracy, this study proposes a motion-induced phase-shifting method that integrates Fourier fringe analysis with speckle correlation-assisted phase matching. In the proposed framework, Fourier transform profilometry is employed to estimate unknown phase shifts in motion-induced three-step phase-shifting sequences and compute the wrapped phase. Phase errors are further mitigated through phase difference analysis and local linear optimization. In addition, a speckle correlation-assisted phase matching strategy is introduced to robustly unwrap the absolute phase for arbitrary surfaces. By utilizing a single random speckle pattern and leveraging triple-view geometric constraints from the stereo cameras and projector, reliable stereo correspondence and absolute phase retrieval are achieved without requiring predefined depth information. Based on the obtained absolute phase and disparity maps, accurate three-dimensional reconstruction is performed using a stereo-structured light system model that fully exploits triple-view information. Experimental results demonstrate that the proposed method effectively suppresses motion-induced artifacts and enables accurate three-dimensional reconstruction under non-uniform motion and non-rigid deformation, offering strong adaptability and broad applicability in dynamic measurement scenarios.