Road pothole extraction method based on improved normal vector distance

Abstract

To address the problem of low accuracy of the traditional 3D laser point cloud pavement potholes extraction algorithm, this paper proposes a road potholes extraction method based on the improved normal vector distance. Then, the normal vector distance of the sampled point is obtained by calculating the distance from the sampled point to the tangent plane of the quadratic surface of the local neighborhood, which is used to describe the 3D features of the sampled point; then, the normal vector distance is diluted and the features are extracted by the Douglas- Peucker algorithm. Finally, the Alpha-Shape algorithm is used to further fit the pit contour and remove the internal noise points, and the B-sample interpolation is used again to fit the boundary of the extracted pit contour to obtain the final pit boundary point cloud collection. The final experimental results show that the average relative error of the pit depth is 3.70%, which is 9.31% higher than that of the traditional method, the average relative error of the extracted pit area is 3.82%, which is 7.33% higher than that of the traditional method, and the average relative error of the extracted pit perimeter is 1.41%. The experimental results show that the method in this paper can extract the pavement pothole features well and improve the performance in terms of the accuracy of pothole feature description compared with the traditional method.