Multi-scale morphological feature differences of coarse aggregates using image processing techniques

Coarse aggregate morphology is closely related to the performance of concrete, asphalt mixtures, and geotechnical materials. To reveal the morphological differences of aggregates from different sources and origins, 900 coarse aggregate particles from natural and mechanically crushed sources were investigated by integrating image recognition technology with statistical analysis and fractal theory. The overall contour features, local angular characteristics, and fractal dimensions of aggregates in the 5–10 mm, 10–20 mm, and 20–40 mm size ranges were systematically quantified. The results indicate that aggregates in the 10–20 mm size range exhibit more complex morphology compared to other size fractions. Of the eight morphological descriptors, seven showed significant differences among aggregate classes (p < 0.05), whereas rectangularity was not significant. Flakiness, shape factor, roughness, and angularity showed the strongest discriminatory ability for distinguishing natural and manufactured aggregates. Principal component analysis indicated that the first two principal components explained 67.29% of the total variance and effectively captured the source-dependent morphological differences of coarse aggregates. In addition, the combination of elongation and flakiness index allows for the classification of coarse aggregates into four distinct particle shapes: flat and elongated, disc-shaped, strip-shaped, and near-spheroidal. Furthermore, sphericity and form factor are categorized into three levels: low, medium, and high. The analysis revealed that natural aggregates were predominantly near-spheroidal, whereas manufactured aggregates were mainly disc-shaped. These findings clarify the multi-scale morphological differences between natural and artificial aggregates and provide a quantitative basis for coarse aggregate selection and engineering application.