Fast Vertical Line Correspondence Between Images for Mobile Robot Localization

The localization of a mobile robot can be estimated by using a single camera attached to it, and by previously knowing the position of a few visual landmarks. This approach is based on the tracking of features over image sequences acquired during robot navigation in indoor environments which, typically, have many vertical lines. In this article, we propose a new and fast method that extracts projections of vertical lines and matches them between consecutive frames. This method is based on a new metric, which considers color distributions in the neighborhoods of the line projections, and on the use of search windows defined by robot displacements. In this way, we obtain an effective and efficient matching process, with empirical evidence that showed promising results.