Mobile Augmented Reality browsers should allow labeling objects A Position Paper for the Augmented Reality on the Web W3C Workshop

Roberto Calvo-Palomino, Juan F. Gato · 2010

LibreGeoSocial is a new FLOSS (Free/Libre Open Source) mobile social network with a Mobile Augmented Reality (MAR) interface that oers functionality beyond the one available on commercial MAR browsers. The nodes of the social network (people, text notes, audio snippets, pictures,. . . ) are both geolocated and registered at dierent altitudes. LibreGeoSocial nodes can be shown through a traditional list based view or through a magic lens MAR UI. This interface allows the user not only to browse tags associated with objects of the reality perceived through the camera, but also enables to use the mobile to link media to objects, enabling the labeling of things at dierent altitudes and situated not necessarily in the vicinity of the user. The system works both outdoors and indoors, in this latest case with location provided by matrix codes. A rst version of LibreGeoSocial was uploaded to the Android market in march 2010. The newest available source code of LibreGeoSocial supports channels/layers that can be updated either from the mobile or through the server. As work in progress, computer vision algorithms have been added to the LibreGeoSocial mobile application to provide tracking of outlines of rectangular objects. With these we intend to explore the best ways to combine the location and orientation provided by GPS and compass with the tracking obtained through image processing in order to improve the accuracy. In this position paper we argue that similar functionality to labeling mechanisms available in LibreGeosocial is a requirement for many kinds of MAR applications yet to come. It is in fact its absence in most current commercial MAR browsers one of the factors that nowadays could hamper the development of this nascent industry beyond the business model exemplied by Layar or Wikitude where users can browse contents uploaded by others third parties.

Read the paper · More papers on PaperTik