Abstract: The estimation of rainfall infiltration is very important in vegetation restoration of slopes. In order to explore the rainfall infiltration process of filling soil under microbial mineralization combined with vegetation in rock desertification slope lattice, based on the traditional Green-Ampt model, the stable infiltration rate is introduced into the surface mineralized soil column, and then an improved G-A model (M-G-A) considering microbial surface mineralization is proposed. For the soil column containing vegetation and considering the infiltration process of composite soil layer, an improved G-A model (P-G-A) considering the effects of vegetation and an improved G-A model (PM-G-A) considering the effects of microbial mineralization combined with vegetation are proposed respectively. The cumulative infiltration is calculated by using the improved G-A model, and the calculated results are compared with the measured values of soil column tests and Hydrus numerical solutions. The results show that the M-G-A model can well describe the variation of cumulative infiltration with time. The calculation accuracy of the model increases with the degree of mineralization, and the error can be controlled within 5% under the action of strong mineralization. The calculation error of the P-G-A model increases with the rainfall duration, but the error can be reduced by about 10% compared with the Hydrus numerical solution. The calculation deviation of the PM-G-A model first increases and then decreases with the rainfall duration, which is generally better than the numerical solution. The improved G-A model can be used to analyze the cumulative infiltration (water-storage capacity) for the ecological restoration technology in microbial mineralized soil in rocky desertificated slope lattice, which has a positive engineering practical value.