Wednesday, January 15, 2014

Augmented Reality Maps using Road Network Tracking

Augmented Reality Maps View using Road Network Tracking
Augmented Reality Maps View for Subway Network
Continuing the experiments using region tracking we have developed in the lab (previous experiment), we tried to apply it for road network. Fortunately, in Google maps, we can make styled maps to make  a road network using unique color that could be easy to extract. Currently, we developed the prototype using handwriting road network like this image.

Handwritten road network for augmented reality maps

We then extract the road network using color extraction method so that we can get the skeleton of the network like this image.
Road network outline for augmented reality maps

Then the next step is finding the descriptor of the curve as explained in the paper of region tracking. Basically, the descriptor is computed using three consecutive points and the value considers the angle that is formed by those points and the length of segments that connect those points. The implementation is pretty straightforward and there is no modification for this kind of outline feature. Then the method track the pattern and the camera pose is computed. Then the annotation of the road network is overlaid in OpenGL.

The following video is the visualization of the AR map using the road network tracking. There are some jitters in some frames due to the lack of descriptors. However, we could show the rotation invariant and the scale invariant by seeing the video.

Next step is applying this technique into printed Google Maps.

No comments :

Post a Comment