Aiming at the problem of ISAR (inverse synthetic aperture radar) super-resolution imaging processing under the condition of low signal-to-noise ratio, an ISAR super-resolution imaging algorithm based on fast split Bregman iteration is proposed. Firstly, in the framework of regularization, the problem of improving azimuth resolution is transformed into a regularization problem. Secondly, the low shift rank feature of Toeplitz matrix and Gohberg–Semencul representation are used to accelerate the convergence. The proposed algorithm not only makes use of the reconstruction ability of split Bregman iteration under the condition of low signal-to-noise ratio, but also ensures fast imaging. A number of experiments are carried out using simulation and real data. The results are compared with the results of existing common algorithms such as LBI (linear Bregman iteration) and OMP (orthogonal matching pursuit). The results show that the proposed algorithm in this paper achieves better imaging performance and relatively short running time.