Lane level accuracy using vision of roadway lights and particle filter

There is provided, in accordance with some embodiments, a method comprising using one or more hardware processors for receiving a stream of video frames from a camera mounted on a moving vehicle. Hardware processor(s) are used for computing two or more three-dimensional (3D) orientation vectors of two or more light sources visible in the video frames. Hardware processor(s) are used for computing a 3D location for each of the light sources based on the 3D orientation vectors. Hardware processor(s) are used for computing two or more geographical locations of the camera based on the 3D locations. Hardware processor(s) are used for computing a lane positioning of the vehicle based on the geographical locations. Hardware processor(s) are used for sending the lane positioning and/or the geographical locations to a navigation system.