Asphalt mixture is a multiphase composite material composed of aggregates, asphalt, fillers and other materials of different properties, in which the coarse aggregate forms the main bearing structure, and the fine aggregate fills the voids formed by the coarse aggregate to improve the structural stability. In this paper, computerized tomography is used to obtain the preliminary tomographic images of asphalt mixture specimens, and the image is effectively segmented through the grayscale thresholding method, and the scanning results are refined. Using voxel-based three-dimensional reconstruction method, the three-dimensional finite element model of asphalt mixture is reconstructed, and the corresponding fine structural characterization index is proposed to prepare asphalt mixture specimens and study the fine structural characteristics of asphalt mixture. The distribution characteristics of the contact connectivity tree of the asphalt mixture are analyzed, with 61.54% of the primary and middle order trees in gradation 1, 92.31% in gradation 2, 83.33% in gradation 3 and 58.82% in gradation 4. It shows that the higher the percentage of second-order connectivity tree, the worse the skeleton contact connectivity, which is not conducive to the improvement of asphalt mixture shear strength. For the four different gradation specimen slice images, the generated areas of each order tree were statistically analyzed. Most of the primary order tree areas were distributed between 100 mm²-300 mm², the intermediate order tree areas were basically distributed between 200 mm²-400 mm², and the high order trees were distributed between 400 mm²-500 mm². The area distribution of high-order tree of grade 4 is more uniform and concentrated, which has better load transfer chain and rutting resistance.