According to the problem that conventional imaging algorithms have some unavoidable shortcomings such as low resolution of image and being fragile to the noise with short coherent processing interval (CPI),an improved ISAR imaging algorithm via CS method,namely,PH-SL0 algorithm,was proposed.In the proposed algorithm,as a kind of measure matrix,partial randomizer Hadamard matrix (PH) has many advantages such as high reconstruction precision and low dimension of measure matrix.Meanwhile,as a reconstruction algorithm,SL0 has many advantages such as demanding fewer measurements than existing methods,having higher reconstructed accuracy and better robust.Therefore,making use of the advantages of PH and SL0,and extending them to the field of radar,the azimuth imaging with short imaging data could be implemented.Finally,simulation results and experimental results of real data show that the algorithm has higher imaging resolution and better robust to noise.