Recently, Dong Shihao, a doctoral student of the Highway School of our school, under the joint guidance of his supervisors Professor Han Sen and Professor Xu Ouming, published a research paper "Asphalt pavement macrotexture reconstruction from monocular image based on deep convolutional neural network.
In this paper, a macro-texture RGB-D data set was constructed through a laser texture and a smartphone, and a 3D macro-texture reconstruction model based on the ResNet-101 residual neural network was established. The RGB image reconstructs its three-dimensional shape. The surface texture reconstructed by the model is compared with the texture scanned by the laser texture instrument, which can accurately reflect the three-dimensional characteristics of the real macroscopic texture of the pavement, and can be used for the calculation of the surface texture evaluation index. To a certain extent, the model can replace the laser texture meter to obtain the road surface texture point cloud data, thereby reducing the test cost and improving the test efficiency, which has important scientific significance for improving the efficiency of road anti-slip evaluation and ensuring driving safety.
"Computer-Aided Civil and Infrastructure Engineering" is a well-known interdisciplinary journal of civil engineering and computer science. The current impact factor of this journal is IF=11.775, ranking first among 137 "engineering, civil engineering" journals, and the Chinese Academy of Sciences and Clarivate Analytics JCR All of the journals in the division belong to the Q1 category of Top journals, and have extremely high influence and reputation in their fields.
