A Novel Iteration Method for Large-Scale Rotation Measurement Using Phase-Based Motion Estimation

The utilization of computer vision in non-contact measurements has garnered substantial attention now, particularly phase-based motion estimation (PME), which stands out for its low computational complexity and high precision in small amplitude motion assessment. However, PME exhibits significant limitations when applied to large-scale rotation measurement. To address this, a novel iterative large-scale rotation measurement method (LR-PME) is proposed in this study. This method breaks large-scale rotation motions into small-amplitude movements frame by frame, dynamically updating the orientation parameter of Gabor wavelet and the coordinates of measurement points. It utilizes Otsu method for threshold evaluation to ensure update accuracy and employs the invariance of pixel distances between measurement points and the rotation center for correction, thereby enabling continuous measurement of large-scale rotation motions. Numerical simulations and laboratory test confirm the robustness of LR-PME, demonstrating superior computational performance.