Geoscience Reference
In-Depth Information
performed in real-time. If we want to use computer vision to
a
location, it is necessary to reduce the set of images to compare. Both these tech-
nologies (feature recognition from CV and GPS) can complement each other, in
order to create a valuable new geolocation method, able to compensate the complete
or partial lack of positioning information.
The main idea of the CV-GPS method is to use available coordinates sent by the
GPS receiver and to start the image analysis process when the GPS signal is weak
or absent, being impossible to accurately reference the current location. To achieve
it, the image analysis process tries to identify a POI in the current captured images.
If one is found, the corresponding GPS location is used. The set of POIs is kept in a
database where, for each POI, one or more images can be kept. Furthermore, in the
database, each image of the POI has the corresponding GPS coordinates associated.
The operation of this method is depicted in Fig. 1 , where X is the limit distance,
in meters, from which the image analysis process starts being executed, and Y is the
maximum time, in milliseconds, that the system may be absent of GPS data before
starting the image analysis process. A system implementing this method receives
and uses the GPS data acquired from the GPS receiver for geolocation. If the GPS
sensor can determine its own accuracy (value in meters correspondent to the
approximate error of the coordinates) then the accuracy information is also used.
geo-reference
Fig. 1 CV-GPS system operation
Search WWH ::




Custom Search