Current research in the field of advanced driver assistance systems (ADAS) sets a focus on vehicle-infrastructure-integration (VII) and cooperative systems. In contrast to state-of-the-art vehicle-limited system, a reliable and accurate positioning becomes a necessary condition for algorithms affecting more than one vehicle. Besides the performance demands of the positioning, also the costs of such a system are a limiting factor. This paper presents one approach for a vehicle positioning system developed in the European Integrated Project SAFESPOT. It integrates GNSS data of an off-the-shelf GPS receiver, standard vehicle odometry, greyscale camera data, and information of a detailed digital map. Therefore, a position estimate obtained from the GNSS/INS system is used to determine the coarse position of the vehicle. Moreover, a two-step Hough transform is used to extract line landmarks like lane markings or curbs from the camera image. Those extracted shapes are then classified using a fast Fourier transform. Finally, the classified landmarks are fused with the coarse position estimate and the detailed landmark data in the digital map to obtain a corrected position.