Radar Automatic Target Recognition (RATR) is the key technique to be breaked through in the fuure development of intelligent weapon system. Compared to the 2-D SAR image target recognition, High Resolution Range Profile (HRRP) target recognition has the advantage of low data dimension, low requirement of radar system's calculation and storage ability, and the imaging algorithm is also not complicated. HRRP imaging is the first and the key process in target recognition, its speed and imaging quality can directly influence the real-time capability and accuracy of target recognition. In this paper a new HRRP imaging algorithm — NUFFT algorithm is proposed, the derivation of mathematical expression is given, both for the echo simulation process and the imaging process. In the meantime, by analyzing each step's calculation complexity, we compared the calculation complexity of four different imaging algorithms, we also simulate two target's imaging and target recognition processing. Theoretical analysis and simulation both prove that the proposed algorithm's calculation complexity is improved in various degree compared with the others, thus can be effectively used in target recognition.