Analyzing asphalt mixture images can provide crucial information not only for generating advanced geometry structure in several numerical computations (i.e. FEM and/or DEM) codes, but also for numerically evaluating the material microstructure. It is well known that 3D X-Ray Computer Tomography (CT) can provide accurate and realistic microstructure information of asphalt mixtures; however, this technology still presents two limitations: 1) the equipment is very expensive and, therefore, only few pavement agencies can afford it, and 2) the software required to generate realistic image of asphalt mixture with three-phase structure (aggregate, asphalt binder and air-voids) is based on a global thresholding algorithm which cannot be easily accessed and edited by users and practitioners.
In this paper, accurate and realistic 2D three-phase asphalt mixture images were generated using an advanced DIP analysis code (implemented on MATLABTM) and a common flatbed scanner, which can be easily purchased at relatively low price. The threshold algorithm was developed based on the computed results of Gmm (maximum specific gravity), Gmb (bulk specific gravity), VMA (voids in mineral aggregates) of given asphalt mixtures which can be experimentally obtained in a laboratory environment. 2D three-phase images of asphalt mixtures were derived from grey scale images (color intensity from 0 to 255) obtained from original RGB (Red-Green-Blue) scale images with a dual-threshold computation techniques (i.e. one step for computing air voids phase, T1, and a second step for computing asphalt binder (and/or mastic) phase, T2). An example of DIP analysis results is shown in Figure 1.
Based on the computation results, quite accurate and good visual agreement was observed between RGB scale image and DIP analyzed image. The findings indicate that this advanced DIP analysis technique can provide reliable geometry and microstructural information for several numerical simulations such as finite element method (FEM) and discrete element modeling (DEM).
This research work represents a solid base for performing simple 2D microstructure analysis using 2- and 3-point correlation function and for developing the Moon Cannone Falchetto (MCF) model which will be introduced in the next annual KSRE conference.

• Augusto Cannone Falchetto(Technical University of Braunschweig (Germany), Post-Doctoral Researcher, Ph.D.)
• KiHoon Moon(Korea Expressway Corporation, Senior Researcher, Ph.D.)