Robust Harris-SURF features for robotic vision based navigation

This paper describes a new alternative procedure to improve the invariance to illumination changes on the image feature extraction provided by Harris corner detectors, using the local image moments information. Another alternative is presented for the matching process, which is enhanced adding local SURF descriptors. These both techniques are joined to give us a new promising visual guidance tool oriented to cope with the visual conditions variations that can occur in the environment of an autonomous robot.