If the source material is images check out Image::OCR::Tesseract, if you have a PDF consisting of one image per page see PDF::OCR, or alternatively PDF::OCR2. All of the above require Tesseract, see Re: parse content of PDF file for a further information.

Martin