ABSTRACT

X-ray micro-computed tomography (micro-CT) is an advanced technique able to provide a comprehensive examination of the volumetric characteristics of asphalt mixtures. A key step for the air void (AV) quantification using micro-CT images is the segmentation, which is a stage of the digital image processing. The most common segmentation technique, the manual threshold (TH) selection, depends significantly on the operator skills, image homogeneity, and material complexity. These factors that can limit the reproducibility of the TH procedure. Machine learning and deep learning recently appeared as promising alternatives to solve this challenge. In this paper, images of an asphalt concrete (AC) specimen were acquired in a modern high-resolution micro-CT scanner to determine its AV content using four different segmentation tools, i.e. TH, watershed, machine learning, and deep learning. All methods presented similar results for the total AV content. The advantages and limitations of using each technique were discussed in terms of computational effort, user-friendliness, and accuracy of the results. Machine learning and deep learning were identified as powerful tools for AC segmentation, being accurate and easy to adjust, however taking longer data processing times.

摘要

X 射线微型计算机断层扫描 (micro-CT) 是一种先进的技术，能够对沥青混合料的体积特性进行全面检查。使用显微 CT 图像进行气隙 (AV) 量化的关键步骤是分割，这是数字图像处理的一个阶段。最常见的分割技术，即手动阈值 (TH) 选择，很大程度上取决于操作员技能、图像同质性和材料复杂性。这些因素会限制 TH 程序的可重复性。机器学习和深度学习最近似乎是解决这一挑战的有希望的替代方案。在本文中，在现代高分辨率微型 CT 扫描仪中获取沥青混凝土 (AC) 试样的图像，以使用四种不同的分割工具确定其 AV 内容，即 TH、分水岭、机器学习和深度学习。对于总 AV 内容，所有方法都呈现出相似的结果。在计算工作量、用户友好性和结果的准确性方面讨论了使用每种技术的优点和局限性。机器学习和深度学习被认为是 AC 分割的强大工具，准确且易于调整，但需要更长的数据处理时间。