mrPCA‐PSP: Robust Motion‐Resistant Phase‐Shifting Profilometry Based on Principal Component Analysis

Abstract

Abstract Phase‐shifting profilometry (PSP) enables high‐resolution 3D shape measurement through phase‐shifted fringe patterns. However, its multi‐shot nature makes it highly susceptible to motion‐induced phase errors, leading to severe reconstruction artifacts in dynamic scenes. Here, a robust motion‐resistant phase‐shifting profilometry method based on principal component analysis (mrPCA‐PSP) is proposed. By leveraging PCA to separate signal from noise, non‐uniform phase shifts induced by motion in each raw fringe image are compensated. Through image rectification and unknown phase shift estimation, the method achieves high‐precision phase retrieval for objects moving in arbitrary directions using only three phase‐shifted fringe images. Experimental results in complex dynamic scenes demonstrate that mrPCA‐PSP exhibits significantly enhanced motion robustness compared to traditional PSP. By implementing it in a GPU‐accelerated multi‐view structured light fringe projection system, motion‐induced errors are effectively suppressed, achieving real‐time, motion‐artifact‐free, high‐precision 3D measurements at speeds of up to 100 Hz. These capabilities position mrPCA‐PSP as a promising solution for high‐speed, real‐time 3D shape measurement in dynamic environments, with applications in industrial inspection, robotic navigation, and human‐computer interaction.