Pavement Potholes Quantification: A Study Based on 3D Point Cloud Analysis

Currently, the detection technology for road surface potholes, primarily focuses on the identification and segmentation, lacking the ability to quantitatively analyze the damage inflicted by road potholes. Therefore, this pa per proposes a method based on three-dimensional point clouds for the identification, segmentation, and reconstruction of road potholes, ultimately leading to the quantification of the damage volume. An RGB-D depth sensor is employed to collect point cloud data of road potholes. Voxel filtering and voxelization downsampling are used for denoising, filtering, and enhancing data processing efficiency. Surface segmentation is achieved through RANSAC (Random Sample Consensus) and Euclidean clustering, while the Alpha Shapes algorithm is utilized for three-dimensional volume reconstruction, facilitating the volumetric quantification of potholes. For evaluation, comparative experiments were conducted under different lighting conditions and shooting distances. The experimental results demonstrate that the proposed algorithm achieves an accuracy of 96.4% in volumetric damage measurement of road potholes, accurately determining the damage volume of pothole.