An innovative approach was developed in this research to generate aggregates based on real shape through DEM. The shape indexes of aggregates were captured by Aggregate Image Measuring System (AIMS). The output was then processed by MATLAB to obtain the edge points of particles. The edge points were used to generate aggregate in DEM. Based on the former procedures, the aggregates morphological database that record shape indexes were established. The generation of DE models was optimized in the study to approach more realistic model geometries. Clump-based models with real morphologies were compared with laboratory tests and conventional ball-based models in the repose angle test. The clump-based model fitted better with the laboratory tests compared with ball-based models. Finally, a pavement compaction model was established. The influence of aggregate gradation and the movement of asphalt mixture during the compaction were investigated.