A Rotation-Matrix-Based Point Cloud Correction Method for Complex Pavement Surface Textures

Accurate reconstruction of complex pavement surface textures is essential for reliable texture characterization and skid resistance evaluation. In this study, a three-dimensional point cloud correction method based on a quaternion rotation matrix was proposed to eliminate point cloud tilting and displacement caused by pavement slope and the initial orientation of the scanning equipment. A unified spatial rotation model was established to align the point cloud normal vector with the Z-axis while preserving the geometric relationships among points. The method was validated using a pavement model with regular sharp textures and then applied to distressed pavements containing cracks, raveling, and rutting. The results showed that the proposed method more accurately reconstructed the original elevation features and effectively avoided the peak–valley blunting and texture smoothing associated with conventional profile-fitting correction. Compared with the conventional method, the average error of mean texture depth (MTD) was reduced by 48.3%, while the error of maximum texture elevation difference (Δh) remained within 39.9%. Statistical analysis indicated that the observed differences mainly resulted from the correction strategy rather than measurement uncertainty. The proposed method improves the accuracy of pavement texture reconstruction and texture parameter calculation and provides a reference for pavement evaluation and digital pavement modeling.