Database Reference
In-Depth Information
Fig. 11.8 The whole MobiTag system
The second step relates the image to a text string. That is, it sends the image of
the object bounded to a Web service that determines what object it is. For example,
in the case of a topic the service will return the title of the topic.
The third step is a Web service that retrieves related information from Internet
and social media related information such as author, genre, criticism, social media
ratings, and prices. It basically finds related information and returns what the
algorithm considers to be most relevant to the user.
The fourth and final step shows the relevant information to the user, by means
of augmented reality. The related information is superimposed on the detected
object. For example, a topic can show an aggregate of ratings in social networks,
an indication of the best price and the option to buy it only by clicking a button or
going to the published information related to the object, etc. All this information is
above the object so that even if the user moves the terminal and if the whole object
appears inside the image, the display information will be provided above, causing
the user to believe that the information is about the object. This is augmented reality.
As the most complex part of the system is to extract the object from the initial
capture and relate this image to with the product's name, we have focused our first
experiments on these initial steps. It was necessary to test the response of CBIR
systems against real images of the object using different illuminations.
The experiment consisted of detecting an object and performing an action
associated with it. CDs were used as objects to be detected and the associated
action was to play a music file related to the captured object. In this case, web
services were not used, only local services such as image recognition and music
launches. This approach simulated the first three steps of the whole system.
The result of the experiment was positive. The preprocessing system (step 1)
improves the CBIR image association by 55%. The percentage of improvement of
Search WWH ::
Custom Search