I have done the same thing a month ago, I used city of Zurich building data-set from ZURICH. In short here is what I have done. the application was a client/server architecture. The client(android phone) graps the view and send the image or SURF/SIFT feature to the server... The server than do the matching and return
I first index the data-set using Bag-of -Words algorithm and associate the name of the building with some other information in the process.
on the mobile phone after capturing the view u can send the JPEG image to the server to recognize and it send back the name of the building back to the android mobile.
in the other option you can process the image to get SIFT/SURF local feature and send this to the server instead of the raw JPEG image.
OpenCV have the implementation of the "bag-of-words" algorithms. It is a the best method to index large set of image for image recognition.
below are 2 links that might be useful
http://www.morethantechnical.com/2011/08/25/a-simple-object-classifier-with-bag-of-words-using-opencv-2-3-w-code/http://opencv.willowgarage.com/wiki/bagoffeaturescheers!
Israel
On Friday, August 17, 2012 1:56:52 PM UTC+2, Christine369 wrote:
I have to build an application which is related to image recognition on Android platform.It should be able to open the camera of the phone and than the building captured in the camera view can be recognized and the name of it will directly show on the screen. But it cannot use GPS to recognize it. Every thing needs to be done by image recognition. What I need is to build a database that relates each building and its name. I am not sure how to build it. Is there anyone who has done a similar task before? Much thanks!!!!
*-*