There is a serious degradation effect on reinforced concrete by chloride ions. Cracks provide a fast channel for diffusion and accelerate the failure of structural durability. In the available publications, scholars focus on studying the influence of a single fracture characteristic factor on the chloride diffusion coefficient, the fitting formula based on the experimental results has limitations. Therefore, in this study, machine vision technology was used to extract and quantify multiple fracture characteristic factors. The representative elementary volume (REV) model of cracked concrete was constructed, and its equivalent diffusion coefficient was calculated by combining the influence of cracking density, interfacial roughness and orientation degree of cracks. Finally, the experimental results in the literature are compared with the calculated results based on the REV model. The conclusions show that the average relative error is 1.6% under different conditions of salt freeze, and 2.9% under different initial fracture conditions after 28 times of salt freeze. The results agree well. This indicates that the proposed chloride ion equivalent diffusion coefficient for cracked concrete has good reliability.