Pavement macroscopic texture reconstruction method based on multi-view deep learning

Abstract

The invention belongs to the field of civil engineering, and discloses a pavement macroscopic texture reconstruction method based on multi-view deep learning. The method comprises the steps: acquiringroad surface multi-view images and road surface macroscopic texture three-dimensional point cloud data as a road surface information data set; establishing a deep learning structure S based on the single-view road surface image for road surface macroscopic texture reconstruction based on the single-view road surface image; establishing a multi-view fusion unit MU for fusing pavement macroscopic texture reconstruction models based on different pavement view images; performing end-to-end training on the deep learning structure S and the multi-view fusion unit MU by using the pavement information data set to obtain a trained multi-view deep learning structure; and generating a macroscopic texture model of the to-be-detected asphalt pavement area by adopting the trained multi-view deep learning structure. According to the method, the three-dimensional pavement macroscopic texture model can be reconstructed by directly utilizing the multi-view pavement images, so that the acquisition of pavement macroscopic texture data is improved, and the method has positive significance for improving the pavement disease and performance detection efficiency.

Description

technical field [0001] The invention belongs to the technical field of civil engineering, and in particular relates to a pavement macro texture reconstruction method based on multi-view deep learning. Background technique [0002] The 3D reconstruction of pavement macro texture is an important part of pavement performance evaluation. At present, the 3D pavement macro texture model is mainly used for the prediction of pavement drainage performance, skid resistance and durability, as well as for the analysis of vehicle fuel consumption, tire noise and vibration. . A good pavement macrotexture is an integral part of a high-quality pavement. [0003] Accurate and efficient 3D reconstruction of pavement macro texture is one of the key points of pavement performance evaluation. At present, the methods for 3D reconstruction of pavement macro texture mainly include handheld grid scanner measurement method and array infrared laser scanning method. However, these test methods are c...

AI Technical Summary

Problems solved by technology

However, these test methods are cumbersome to operate, have high technical requirements, and low detection efficiency. The detection process needs to close the traffic, which seriously affects the efficiency of pavement performance evaluation and affects the efficiency of road traffic. It is not conducive to the large-scale promotion of pavement detection in my country.

[0004] Multi-view pavement images can accurately characterize pavement macro texture; however, due to the complexity and diversity of pavement materials and image acquisition conditions, multi-view pavement images are difficult to directly use for 3D reconstruction of pavement macro texture

Images