One factor of increased violations on the highway is a violation of traffic signs, because the signs are not visible to the driver. In addition to this the conditions on road signs attached to the road have shortcomings such as twisting signs, imperfect beacons and non-standard beacons. So to be able to reduce the violation of traffic signs required a system that can recognize traffic. To be able to recognize traffic signs can be done visually and must be fast in recognizing. The image to be recognized can use the camera to retrieve information from signposts then the image is extracted with features of Speeded Up Robust Features (SURF) algorithm consisting of three stages: interest point detection, scale space, feature description and feature matching so that the system can recognize traffic signs. The research that has been done has resulted that the SURF algorithm in recognizing traffic signs is about 83,33% accurate to be the algorithm of introduction of traffic signs with the need to be fast and accurate. In addition, this algorithm is invariant to scale and invariant to rotation, so that the difference of slope and scale difference can still be recognized by using SURF algorithm.