The compaction process of pavement during construction critically influences the mechanical performance of the pavement. However, this sophisticated procedure, including asphalt mixing, pre-paving, and multi-step compactions, could induce significant compaction variation, which, to some extent, might cause unwanted distresses in the early service of pavement. An insight into the in-situ compaction variation would benefit the estimation of compaction quality and further optimize the compaction procedure. For this reason, this study estimated the in-situ compaction quality incorporated with an uncertainty analysis. A total of 45 samples were cored from a newly constructed test track. The volumetric and microstructural parameters (air void content, degree of compaction, fractal dimensions of aggregate and air voids) were calculated through digital image processing for each sample. Additionally, the mechanical property (indirect tensile strength) of the asphalt samples was measured by indirect tensile test. Subsequently, two uncertainty approaches, namely the non-probability convex model and decision tree, were employed to feature the data distribution and figure out the priority factor that influences the mechanical performance of asphalt mixtures. The results indicated that the microstructure of asphalt mixtures remarkably affected the distribution of the degree of compaction. Additionally, asphalt mixtures' indirect tensile strength was associated with compaction indicators differently depending on the measuring temperatures.