Construction and optimization of asphalt pavement texture characterization model based on binocular vision and deep learning
Document Type
Article
Publication Date
5-15-2025
Department
Department of Civil, Environmental, and Geospatial Engineering
Abstract
To efficiently characterize the texture of asphalt pavements, a BCTD (Binocular Camera Texture Detection) system is developed based on the principles of binocular stereo vision technology. The system introduces an innovative approach to texture analysis using self-invented TDFA (Tire-pavement Dynamic Friction Analyzer) equipment and long-term anti-skid performance testing. The system facilitates the collection and pre-processing of pavement texture images, achieving an average TOP1 accuracy of 67 % with a measurement precision of 0.1 mm. The results indicate that the model exhibits excellent recognition performance for weak feature images within the same type of pavement texture, effectively characterizing the texture of asphalt pavements. In summary, this study provides a comprehensive and innovative approach to asphalt pavement texture characterization. It advances the field by providing valuable insights into texture analysis, particularly weak features. The BCTD system demonstrates the potential of monitoring the skid resistance of asphalt pavements to improve road safety and maintenance efficiency.
Publication Title
Measurement: Journal of the International Measurement Confederation
Recommended Citation
Yu, M.,
Zhang, R.,
Tang, O.,
Jin, D.,
You, Z.,
&
Zhang, Z.
(2025).
Construction and optimization of asphalt pavement texture characterization model based on binocular vision and deep learning.
Measurement: Journal of the International Measurement Confederation,
248.
http://doi.org/10.1016/j.measurement.2025.116946
Retrieved from: https://digitalcommons.mtu.edu/michigantech-p2/1432
