OCR returning wrong result

Hi

I'm running Sikuli 1.1.1 on Windows 7.

In this script (made as simple as possible to reproduce the issue), I try to find a text displayed as an image :

reg=find("1491816672482.png")
found = reg.text()
print found

The image contains only a very readable "10841" in times new roman font , colored red, background light grey, sized 20 pixels. (can't I attach a screen copy here ?).

but found is only the string 1084 : I'm missing the the final "1" and this lead to wrong result of the test !

My questions :
  - Are there options to improve the OCR, or anything I could check or change to allow correct recognition ?
  - Is this font known as difficult to read by OCR ?

Thanks