Imaging objects in electrical impedance tomography (EIT) measurement are often in a dynamic evolution process, and exploiting the spatial-temporal properties for the dynamic reconstruction objects is crucial for the improvement of the reconstruction quality. In this paper, the linearized equation system for dynamic EIT is established based on spatio-temporal representation. Due to the spatial redundancy and the temporal correlations, the low-rank property of a two-dimensional difference image at a specific time window is introduced to objective functional. The Augmented Lagrange Multiplier (ALM) method is applied to solve the new objective functional. The effectiveness of the proposed scheme is illustrated on both simulation and experiment data sets.