Feature-based alignment method for projecting virtual content on a movable paper map

In this paper we propose a feature-based alignment method for visualizing geographical content on a movable paper map using non-calibrated projector-camera pair. In order to project correctly aligned geographical content on a movable paper map, we detect two kinds of features: printed features and projected features on a paper map. First, we print landmark points as map features on the paper map for registration purpose. We then project landmark points onto the paper map. We then detect these printed and projected features simultaneously. The transformation from the projector coordinate system to the image coordinate system is estimated using the feature correspondences. This transformation is used to warp the geographical content before the corrected projection. In succeeding frames, the features are tracked in order to update the transformation. We evaluated our method against the motion of the paper map, the camera and the projector with average projection error of 9.5 pixels.