Evaluation of Two-Dimensional Gray-Scale Images for Microtexture Analysis of Aggregate Surface

Abstract

A powerful texture feature extraction algorithm based on transform-based analysis of two-dimensional (2D) gray-scale images was employed in the aggregate image measurement system (AIMS) to quantitatively describe aggregate surface microtexture by way of a parameter called texture index (TI). TI has been used successfully to distinguish between unpolished and polished aggregates. The main goal of this research was to further evaluate this system using a broader range of aggregates in terms of mineral composition including highly uniform limestone sources and highly variable granitic sources. Results confirmed that TI successfully distinguished polished from unpolished aggregates for all mineral compositions. However, an excessively broad range of TI values was observed. Specifically, TI values of aggregates with highly variable mineral composition were far greater than any value previously reported in the literature. Interestingly, unusually high TI values were also observed for limestone aggregate particles obtained from field cores. It was hypothesized that non-roughness-related features such as surface color pattern resulting from mineral variation and absorbed asphalt resulted in artificially high TI. Independent experimental roughness evaluation using scanning electron microscopy (SEM) and surfaces prepared to the same roughness level confirmed this hypothesis. The results clearly indicated the need for a change in image acquisition and/or analysis algorithm to exclude the effect of surface color pattern in texture analysis of aggregate surface images.