In order to separate adhesive aggregate particles of asphalt mixture in CT(computerized tomography)images, based on the continuous density of the materials, the Gaussian smooth and H maxima transformation for CT images were performed to eliminate the noise and decrease the pixel density variability of aggregate particles in image, and then the watershed segmentation algorithm was used to segment the adhesive aggregate particles. The size parameters of aggregate particles segmented by watershed were determined. Furthermore, gradation of aggregate particles was virtually screened using different equivalent sizes, the method for segmenting adhesive aggregate particles was verified using screening gradation. The results show that the equivalent diameter can be regarded as equivalent size to conduct virtual gradation screening for aggregate particles segmented by watershed. The virtual gradation screening result for aggregate particles has good correlation with the actual gradation screening results. This method overcomes the defects of previous research ignoring to process the adhesive aggregate particles in CT images. It is a useful exploration for designing gradation of asphalt mixture digitally.