Molding granule is the bridge between Explosive crystal materials and PBX. Quantitative characterization of their physical structure is the key to achieving the performance regulation and improvement of PBX. Based on micro-CT， image processing， and statistical analysis is researched to quantitatively obtain the characteristic parameters of molding granules. Image processing and statistical analysis were researched to quantitatively obtain the characteristic parameters of molding granules based on micro-CT. The characteristic parameters of TATB-based molding granules were characterized by this method， and an evaluation of the characterization method was carried out. The results suggest that this method can implement the quantitative characterization of the size， morphology， porosity， density， and packing characteristics of the molding granules. The granule size and sphericity of the TATB-based molding granules approximately obey the exponential Gaussian and Weibull distributions， respectively， and the intrinsic density is 1.35 g·cm^-3， visible porosity of 2.3% （@ spatial resolution 14.3 μm）， volume fraction of 0.69. Evaluation experiments show that the method has good accuracy， stability， and repeatability， the characterization results of granule size are consistent with the results of the sieving process， and the relative deviation of the characterization results of main characteristic parameters in different periods are less than 1%. The results of morphology and properties are relatively greatly affected by the packing state of granules and the spatial resolution of micro-CT.