I am trying to build a system (using C#) that can recognize text for scene images. I see that scene text recognition is a challenging task because of low resolution, complex background, non-uniform lightning or blurring effects...
Any ideas for overcoming this problem would be appreciated.