Feature representation and matching are two challenging problems for person e-identification problem. Designing a suitable feature representation method, and the according high efficiency matching scheme is meaningful. In this paper, a new person re-identification method was put forward. First, an improved BOF method was proposed, it use SURF algorithm to extract the preliminary feature and generate visual dictionary. And then, an effective classifier was designed by LIBSVM technology. This method can deal with illumination and scale invariants, and improve the efficiency of the matching process. The experimental results show that it is an effective scheme for person re-identification.