基于数字图像技术的沥青路面纹理重构和抗滑性能评价方法综述

Review of Asphalt Pavement Texture Reconstruction and Skid Resistance Evaluation Based on Digital Image Technology

摘要:

沥青路面的表面纹理特性直接决定其抗滑性能，并影响道路交通安全。传统的沥青路面纹理测量方法（如铺砂法、环形纹理测试仪法等）在测量精度、操作便捷性及环境适应性方面存在一定局限。近年来，随着计算机视觉与图像处理技术的发展，基于数字图像技术的沥青路面纹理重构与抗滑性能评价方法逐渐成为研究热点。系统综述了数字图像技术在沥青路面纹理表征中的应用，重点探讨了4类常见的纹理重构方法，包括基于灰度图像、立体视觉、光度立体视觉及三维激光视觉的重构技术，并分析了各自的优缺点及适用范围。此外，梳理了4种主要的沥青抗滑性能评价方法，包括基于几何学指标、频谱分析、分形理论及人工智能的评价策略，归纳了相关研究成果及其在不同工况下的适用性。研究表明：数字图像技术能够精准重构沥青路面的三维纹理形貌，并通过提取关键特征参数（如构造深度、轮廓曲率、分形维数等）实现对抗滑性能的科学评价。深度学习和机器学习等人工智能技术的引入，使得基于图像数据的抗滑性能预测达到了更高的精度和自动化水平。尽管数字图像技术在实验室和现场检测中表现出较好的适用性，但仍面临计算复杂程度高、环境适应性不足及工程应用标准化程度较低等挑战。未来研究应重点关注多模态数据融合、智能算法优化、实时监测及标准化应用，以进一步提高路面抗滑性能评价的精度和工程实用性，为道路安全和智能交通系统管理提供科技支撑。

Abstract:

The surface texture characteristics of asphalt pavement directly determine the skid resistance and affect the road traffic safety. The traditional pavement texture measurement methods, such as the sanding method and circular texture meter, have certain limitations in measurement accuracy, operational convenience and environmental adaptability. In recent years, with advancements in computer vision and image processing technology, the pavement texture reconstruction and skid resistance evaluation based on digital imaging technology have gradually become the research hot spots. The application of digital image technology in asphalt pavement texture characterization is systematically reviewed, and the focus is on four common texture reconstruction methods such as grayscale image-based, stereo vision-based, photometric stereo-based and 3D laser vision-based techniques. The respective advantages, disadvantages and applicability are analyzed. Additionally, four main skid resistance evaluation methods are sorted out, including the evaluation strategies based on geometry index, spectral analysis, fractal theory and artificial intelligence. The relevant research findings and the applicability under different conditions are concluded. The study results show that digital imaging technology can accurately reconstruct the 3D texture shape of asphalt pavement and scientifically assess the skid resistance by extracting the key feature parameters, such as texture depth, contour curvature and fractal dimension.The introduction of AI techniques, such as deep learning and machine learning, can make the skid resistance prediction based on image data achieve a higher accuracy and automation level. Although the digital image technology demonstrates the good applicability in both laboratory and field detection, it still faces challenges such as high computational complexity, limited environmental adaptability, and a low degree of standardization in engineering applications. Future research should focus on the multimodal data integration, intelligent algorithm optimization, real-time detection and standardization application to further enhance the accuracy and engineering practicality of pavement skid resistance evaluation, which provides the scientific support for road safety and intelligent transportation systems management.