To effectively evaluate the construction homogeneity of asphalt pavement, the tomography image of a core sample of asphalt pavement was obtained via industrial computed tomography (CT) equipment. According to the characteristics of CT images, an improved separation algorithm based on annular partition and Nobuyuki Otsu (OTSU) threshold segmentation was proposed. Based on the distribution of aggregates, voids and asphalt mortar, and the area ratio of each part in the CT images inside the pavement, a novel evaluation method for the distribution homogeneity of asphalt pavement components was put forward, and the validity of the evaluation index was also verified. The results show that the aggregates, voids and asphalt mortar in CT images can be effectively segmented by annular partition combined with the OTSU threshold separation algorithm. By superimposing the segmented image on the original image, the segmentation and identification effects of aggregates, voids and asphalt mortar in the CT image are confirmed. Compared with a non-segregated specimen, the average values of the horizontal heterogeneity coefficients of high, medium, light and fine-aggregate-segregated mixtures increased by 72.0%, 48.3%, 34.7% and 16.1%, respectively, where the change range is in accordance with the segregation degrees of several mixtures. The indirect tensile strength of fine-aggregate-, light-, medium- and high-segregated asphalt mixtures decreased by 8.3%, 16.7%, 25.0% and 45.8%, respectively, when compared with the non-segregated asphalt mixture. The index of the vertical heterogeneity coefficient has good correlation with the indirect tensile strength of segregated asphalt mixtures. The construction quality homogeneity of asphalt pavement in different regions can be reliably evaluated by the horizontal heterogeneity coefficient and vertical heterogeneity coefficient.