Visual localisation

A method of visual localization, comprising:

• obtaining, during a mapping phase, using at least a first camera, one or more real reference images of an environment, wherein each of the reference images is associated with a viewpoint;
• pre-computing, after the mapping phase, using image based rendering, a plurality of second images or image patches, each of the second images or image patches being generated by warping one or more of said reference images, or at least one image patch from said reference images, to simulate a view from a secondary viewpoint, wherein each of the secondary viewpoints is associated with a location determined based on a floorplan of the environment;
• extracting features from said pre-computed second images or image patches and combining said extracted features into a document that is indexed by a content-based image retrieval (CBIR) system;
• obtaining at least one query image;
• and,
• determining a location, utilizing said document, of the query image on the basis of a comparison of the query image with said pre-computed second images or image patches.