To address the problem associated with the relatively low temporal resolution of the multi-axis differential optical-absorption spectroscopy used for reconstructing images, we introduce a method to reconstruct the two-dimensional spatial distribution of SO2 gas by combining the imaging differential absorption spectroscopy technology with the non-negative least square method. The sampled column density data are preprocessed using a Savitzky-Golay filter, the sequential coordinate-wise algorithm is used for reconstruction, and Kriging interpolation is used to smooth the reconstructed image. The numerical simulation results demonstrate that the nearness index of the reconstructed image is the lowest (0.11). The experimental results show that the time required to collect data and reconstruct an image using the proposed method is 52.37 s, which is applicable to the capture of the transient concentration of a plume cross section and the real-time reconstruction of SO2 distribution.