Acrobat pro 9 is what I use... 300dpi scans and above generate spot on OCR with few errors, go recognize text with OCR. Then you can export it as txt/rtf/html (if it's just a simple book)
There's other software out there too, but if it's already a pdf, shrug.
New version of Nitro PDF has OCR so I've heard. There also may be other free software out there, not sure about the quality. Some more advanced OCR software I have used in the past allows fine tuning of OCR and imports of dictionaries and such. I think acrobat just uses standard dictionary for OCR correction, and it doesn't ask you to correct errors if it finds any
Good luck to you.