My project for this summer is to build a cell phone application that
will allow users to translate or hear aloud any text of which they've
taken a picture using their phone's camera. We expect this application
to aid many kinds of users: those learning to read, those learning a
new language, those living in a country whose language they
don't speak, those who think hearing everyday phrases in different
languages is funny... And the list goes on.
This application will be developed using Android. It may be cross-developed on another platform for comparison. The reasons behind this design goal are to familiarize myself with a recent development environment and to communicate my experiences working with Android such that it may be used in Dr. Martonosi's lab in the future.
As it turned out, my project became more of an exploration of image preprocessing - in order to translate text from images, you first have to recognize the text characters with an OCR program, but it's hard for available OCR programs to recognize text in a complex background. If you look over at my final report and the code I've posted, you can see more about my attempts to produce from natural scene images "clean" images of only text that OCR engines are better able to recognize.
|My Final Report