In general,industrial CT images of asphalt mixture are processed via the threshold segmentation to obtain the global threshold for the separation of aggregates from other materials.In order to presents an effective method that comprehensively evaluates the image segmentation of asphalt mixture,the characteristics of image gray histogram distribution are considered,and the Gaussian mixture model（GMM） and the fuzzy C-means（FCM） clustering algorithm for pattern recognition are introduced in the image process.Then,a two-classification strategy is adopted twice to divide the image into four different regions respectively corresponding to background,voids,mastics and aggregates.Moreover,a reference material is produced to calibrate the CT results,and three methods for material classification are used as the auxiliaries for the valuing of reference material to comprehensively compare the computational efficiency and the segmentation effect.The results show that the OTSU method is more efficient than GMM and FCM for the processing of images with clear double-peak distribution because it has no need for improvement.