A Digital Evaluation Approach of Asphalt Pavement Mean Texture Depth from 3D Point Clouds

Abstract

To address the low efficiency and operator dependence inherent in traditional methods for measuring asphalt pavement mean texture depth (MTD), this paper proposes a digital evaluation method based on three-dimensional (3D) point-cloud data. Pavement point clouds are acquired using a self-developed binocular vision system in conjunction with a circular hollow physical mask. A preprocessing algorithm integrating random sample consensus (RANSAC) and adaptive hybrid filtering is employed for pose correction and denoising. The point-cloud region of interest (ROI) is divided into equal-area circular sub-blocks, from which local peak-points are extracted and used to construct a reference surface. MTD is subsequently calculated using an infinitesimal-volume accumulation approach. Comparative experiments are conducted to determine the optimal sub-block size and fitting model for this dataset. Results show that the combination of a 7 mm nominal equivalent sub-block size and the thin-plate spline (TPS) model achieves a Pearson correlation coefficient of 0.977 relative to the standard sand patch method (SPM). Furthermore, given identical point-cloud inputs, the proposed scheme shows better overall agreement with the SPM than the alternative calculation approaches adapted in this study. The proposed method provides a digital macrotexture assessment tool; its implications for skid-resistance prediction require further independent validation. The 7 mm setting should be regarded as the best parameter for the present dataset rather than a universally established optimum.