Localization based on vision using stereo System and SIFT features

Abstract

The traditional method uses the GPS signal to determine the location of a vehicle or Mobile Robot, but these methods sometimes fail to determine the exact position of the object especially in urban areas, or inside buildings. It fails to determine the position for multiple reasons such as the multi-fading path, obstacles, rain, and snow which all have an effect on the signal of the GPS. The developed algorithm is intended to get the location of the moving objects by using the computer vision system. The results of the proposed algorithm give higher performance (around 33 second and with a resolution of 10 meters) compared to the other systems on the long trajectory. We use the single camera system to select the suitable features and the stereo camera systems to obtain the locations of the object. It is therefore different from the other methods that are dependent on maps to compare the image features with the features of the map to determine the location of the object.