Optical flow computation for land-based mobile mapping system images

Abstract

The positioning of a land-based mobile mapping system (LBMMS) is obtained by the GPS signal processing. However, buildings and trees, for instance, can cause obstruction to the GPS signal reception, while the vehicle travels along a road. There are mobile systems that have inertial sensors integrated with GPS receivers to orient the road images. Such integration can not control the error propagation in the inertial mechanism so that the image orientation is not possible so that the images result misoriented during the lack of the GPS signal. An alternative to external sensors is the image motion analysis through the binocular optical flow. This article presents the optical flow computation by the differential method in the context of the road images acquired by the digital stereo video camera of a LBMMS. Many tests were performed combining the image quantity of a sequence and the optical flow regularization parameter in order to determine the parameters to be used in such situation. Maybe in the future the optical flow combined with photogrammetric and navigational techniques can contribute to the photogrammetric traverse and image navigation.

Keywords

• photogrammetric traverse
• mobile mapping
• optical flow
• image sequence.