Image-Based Evaluation Method for the Shape Quality of Stacked Aggregates

Coarse aggregate shape plays a critical role in determining surface performance and durability in pavement systems. Traditional manual shape inspection is laborious and subjective, especially for bulk aggregates in overlapped state. In this work, we propose an automated digital image-based evaluation method for stacked coarse aggregates, combining preprocessing (grayscale conversion, histogram equalization, Gaussian filtering), segmentation, and contour reconstruction via the Graham scan convex hull algorithm. Morphological parameters such as equivalent ellipse major/minor axes, area, and perimeter are then extracted to compute individual particle shape factors. To assess batch-level quality, shape factor standard deviations (σ) and mean shape factors were computed from 50 aggregate images. Comparison with manual measurement results shows mean relative errors below 15%. Our analysis reveals a strong correlation between σ and overall shape quality: lower σ indicates more uniform geometry, while higher σ suggests greater irregularity. Based on experimental data, we define three σ-based categories: excellent (σ ≤ 0.32), good (0.32 < σ ≤ 0.42), and poor (σ > 0.42). This σ-driven evaluation framework enables rapid, quantitative, and objective assessment of aggregate morphology in practical aggregate production and pavement quality control.