I don't think the format itself (PDF vs LRF) matters much if it's scanned text. The more important thing would be to remove the whitespace, and divide the columns to get readable output, so I wouldn't just convert it in Calibre or similar, but actually do some processing of it.
I would feed it through
PDFLRF or similar program personally (
Rasterfarian,
PDFread,
SoPDF,
PaperCrop... or if you have the time, cut and paste the exact columns of text as separate pages using Acrobat).
I've had the best kind of luck with PDFLRF for this kind of job, personally. Unfortunately, the link for it is no longer available here at Mobileread, but you can find it if you search hard enough for it.
Of course, there's no reason to try it just as is, and see how well the reflow works, but you might not like the results.