![](https://seccdn.libravatar.org/avatar/ae9e1c52f7693f8923e040894660e5e9.jpg?s=120&d=mm&r=g)
On Sat, 2004-11-20 at 10:20, John Pettigrew wrote:
In a previous message, Mike McMullin
wrote: I am looking for a good OCR package for 9.1, I do not have an installed scanner. Does anyone have any suggestions or experience that could be helpful?
I use kooka to control gOCR, although ocrad is also available. I do it this way because I find it easier to control the parameters and because it has dictionary checking available, which increases accuracy.
I ran an image through it, and it managed to save the "corrected" text. I think I missed something.
However, OCR results are extremely dependent on the quality of the input scan image, so not having a scanner yourself (and thus having no control over the input quality) will make life rather difficult. If you're planning to do much OCR, you could make your life a lot easier by getting a reasonably cheap scanner - almost anything new would be good enough for OCR; the higher prices mostly get you better colour response, which isn't relevant for OCR.
I have an HP scanjet3670, which is windows only, so if I want to dual boot I can control the scan quality. I've been tossing around the idea of installing the scanner's windows software under wine and see if that works at all. (I need to find it first of course. :( ) Thanks, Mike