This paper addresses the problem of self-location for a mobile robot equipped with a single camera moving in an indoor environment. The robot is provided with a two-dimensional map where the position and some attributes of landmark points are stored. The proposed algorithm first determines the observation rays of vertical edges extracted from one image and then finds an interpretation for these rays in terms of the landmark points. This interpretation is driven by a set-based approach that compels the actual pose to lie in a solution region and not to violate the landmark attributes. Based on the ray-landmark matches provided by the selected interpretation, an optimization procedure is used to come up with the pose for which the mean square angular error is minimum. Finally, we present experimental results that demonstrate the performance of the system.