Establishing　a　mesoscale　model　of　asphalt　concrete　is　significantly　challenging　due　to　its　inherent　heterogeneity　and　high　proportion　of　aggregates.　Initially,　a　novel　approach　for　systematically　quantifying　aggregate　morphology　by　integrating　both　form　scaling　and　spherical　harmonic　(SH)　modeling　is　formulated.　The　proposed　method　excels　in　decomposing　aggregate　morphology　at　diverse　length　scales　and　accurately　quantifying　non-star-like　and　flat　aggregates.　Subsequently,　Principal　Component　Analysis　(PCA)　is　performed　on　the　quantified　morphology　parameters　to　produce　sufficient　virtual　aggregates　with　similar　morphology　to　the　real　ones.　Finally,　the　robust　Bullet　physics　engine　is　employed　to　compact　the　generated　aggregates,　which　can　develop　an　aggregate　packing　structure　closely　resembling　real　asphalt　concrete.　Besides,　the　aggregate　morphology　and　gradation　of　the　generated　packing　structure　can　be　effectively　controlled.　Through　additional　geometry　and　mesh　processing,　high-fidelity　mesoscale　models　of　asphalt　concrete　can　be　generated.　This　novel　strategy　lays　the　foundation　for　further　mechanical　modeling　on　asphalt　concrete.　©　2024　Elsevier　Ltd