Answers are grounded in this paper and cite their sources.
Three-dimensional (3D) laser scanning technology is widely used in road detection, where accurate point cloud data serve as the foundation for detection and analysis applications. To address noise and tilt artifacts in high-precision 3D point cloud raw data, this paper proposes a preprocessing method for asphalt pavement point clouds by integrating wavelet and variational mode decomposition (VMD). First, high-precision 3D point cloud data of asphalt pavement were acquired from different sections of an urban greenway using an LS-40 pavement analyzer. The 3D point cloud data were then decomposed into 2,048 two-dimensional (2D) cross-sectional profiles. VMD was applied to decompose each profile into multiple intrinsic mode functions (IMFs), and noise-dominant IMFs were identified and denoised using wavelet decomposition. The warping effect was eliminated by subtracting a quadratic polynomial-fitted elevation tilt term from the original elevation data, completing the preprocessing. Finally, the denoised 2D profiles were reconstructed into a 3D point cloud. The preprocessing performance was evaluated using the Kurtosis metric and subjective visual quality assessment. Results demonstrate that the proposed method more effectively removes noise and tilt trends compared to mean, median, and Gaussian filtering.
No takes yet. Share an insight, caveat, or question.
Ding et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: