Aimed at the problem that by using the polar format algorithm (PFA) the computation amount is large and its imaging precision is affected by interpolation, a fast polar format imaging method of MIMO radar with sparse array based on two-dimensional CZT is proposed in this paper coupled with compressed sensing (CS) and the Chirp-Z translation (CZT). Instead of the interpolation, the CZT achieves the interpolation-re-sampling process by selecting proper parameters. With the imaging size changing, the simulation results show that the amount rate of computation work by the proposed algorithm to the PFA is maintained between 0.22～0.4, thus reducing the computation amount and improving the precision availably with small image entropy.