To investigate the influence of background atmospheres on detection stability and optimize the signal performance of LIBS for soil analysis, plasma imaging and numerical fitting were employed to conduct a qualitative analysis of the laser-induced plasma evolution of strontium in soil under four types of atmospheres, namely air, N?, O? and CO?. Based on imaging of the plasma morphology evolution process using an sCMOS camera, it was observed that in N2 background atmosphere, the plasma exhibited a slower decay rate of electron density, with the most pronounced luminescence profile. This indicates that the plasma stability and sensitivity are optimal in this environment. The changes in electron density and plasma temperature over time are displayed through numerical fitting. The results indicate that the trends of electron density and plasma temperature in the time domain follow an exponential decay pattern. In N2 background atmosphere, the electron density and plasma temperature reached their maximum values, and their decay rate was slower compared to other gases, which is consistent with the evolution process of plasma morphology. The quantitative results of Sr in the soil show that the maximum fitting coefficient (R2) in N2 background atmosphere is 0.97792, with the overall smallest relative error (RE) in concentration inversion. The root mean square error (RMSE) reached a minimum value of 0.178 mg/kg, and the limit of detection (LOD) was 41.90 mg/kg, achieving optimal detection performance. This study will provide insight into the analysis of the plasma evolution process in LIBS technology.