I found Tesseract OCR worked well on my test image if I re-sized the
image to 5 times its original size.
I guess that since Ocropus is in a development stage it is not yet
suitable.
> I tried it out on this imagehttp://
orsa-studio.dyndns.org:81/a.gif
>
> ocropus page a.png
> Annie Ha2
> [beam search failed]
> [beam search failed]
> no6fms 93 mmotes
> [beam search failed]
> [beam search failed]
> [beam search failed]
>
> I tried the same image with a commercial OCR and it worked well.
> There's no way I can afford to pay them.
https://www.ocrterminal.com/