Android Interface

For a while we are studying on fancy parts of android programming. For example, we have tried to do open related browser in application. We had problems with that because we knew nothing about android programming. But after searching we learnt how we solved our problem by using thread. Then we are starting to do some fancy parts.

At the begining we put exit button to terminate application.

Then we redirect user related web page in application. And we put back button to go back and search for different pages.

Addition to this, we give 7 seconds to application to search content. After that it returns main RGB status to search different page.
Second thing that I want to mention that Enlarging database is caused some problems. Because we had 12 pages totally. And some pages does not have certain features so matching score returns low level. Now we had 4 images. But we are trying to add different images too. But it is long and difficult process. Because to prevent mismatching, we have to test all images and sometimes updates our treshold value. Now we are working on building application which will become faster and give best, correct result.

Enlarging Database and Exracting Descriptors at Ones

Until now,we executed our application according to  one picture. When we decided to enlarge our image database, we encountered lots of problems. The larger database we have, the more modular code we have to write. With this approach:

•  We detected key points at the beginning of the program execution ,
•  extracted descriptors from these keypoints
• and then we had descriptor list of all our images.

While program is running, takes the scenes frame by frame, extracts descriptors then compares these descriptors with our image descriptor list. After getting the best good match score, it returns related image location. Then application redirects user to the related image's content web link.

At this point we have three different images.Two of them are unrelated with the scene which we direct our tablet cameras. And one of them is true picture of the camera vision. And our application can distinguish our scene from other two unrelated pictures and redirects us to related web sites. And now application is faster than the way it was before although now it has more pictures to work.

Redirecting To Related Content

Our project's main purpose is when good matching score is obtained user will be redirected to a content. To accomplish this we made button  "Find Features" for matching process.

When user pushes the find features button, matching process starts and matching score writes on screen.

After matching score passes the threshold value user is redirected to "www.google.com".