Explanation for a tesseract result

[sreeragh AR]

I am working with tesseract 3.05.01 (stable version). 
I am using to recognize single charcter images.
It is working fine in most of the cases (with blurred or even noisy images).

But can't understand why it fails in a reasonably clean image.
The image below is identified as 'E' instead  of '0'.

Tried some image processing using OpenCV like binarisation, deskew, denoising etc. 
Still same result.