Aiming at the issue of object location in 3D space, we proposed a method to realize binocular visual positioning with single camera, which is simple, convenient in operation and cost-effective. We used the SURF algorithm for image feature point extracting and matching in the process of object recognition and locating, since the algorithm has good performance in all respects. In order to reduce error matching in the matching process, improvement was made to the SURF algorithm. The experimental result indicate that, the method of monocular to binocular visual positioning based on improved SURF algorithm has fine feasibility and practicability both in accuracy of positioning and in time cost, which is of a certain practical value.