OCR always crashes after some 150 or 200 pages!

Question

Acrobat Pro (I'm using version 10.1.1 currently, but the problem I am identifying here has been around since at least version 8.0, if not earlier) does a fine job of OCR-ing my documents which contain page after page of 600 dpi text data; but if the document is longer than about 200 pages, the application invariably crashes, leaving me with nothing---a terrible design, if you ask me. For starters, why can't the output file, from time to time, be saved to disk so that if a crash happens not all is lost? On long documents, the problem is especially aggravating because the process has by that time consumed quiet a lot of time, usually 40 minutes or more.
But, of course, the application should not be crashing in the first place. I've noticed that the speed at which the individual tasks occur for each page (loading image, deskewing, rotating, decomposing, OCR-ing, image processing, post processing, and generating output) each take longer and longer to perform as the "Recognize Text" request makes it way through the document. I suspect there is in addition to an implementation that fails on large documents also an accumulative memory leak (but of course I can't be sure of that), or at least a demand that more and more memory is required for each successive processed page. At any rate, all I know is that application bombs out. I have, for years, always permitted the resulting crash/debug data that Apple throws up to be sent to Adobe, but nothing ever seems to come of it. Is anyone else having this problem?
The bulk of what I OCR is comprised of Google Book data that I have downloaded. Google provides their book data at 600 dpi. I've tried downsampling to 300 to reduce the output size (which balloons dramatically when the file is OCR'd; or perhaps that's due to Adobe's resampling algorithm, I can't be sure), but this does not solve the problem---and by this I mean that the problem persists whether the file is first "optimized" (that is, with settings for grayscale images to be "downsampled" to 300 dpi using the "bicubic" method), then saved, and then reopened for the OCR step, or whether I request images to be simultaneously downsampled to 300 dpi during the OCR step itself. Unfortunately, reductions below 300 dpi result in too much degradation of the text and dirtying of the background, so I cannot re-render the files in a resolution smaller than 300.
My only solution thus far is to split the downloaded books into parts of about 150 pages or less--but what a pain this is, as almost every title I wish to OCR is far larger than that!
Any suggestions for an alternative methodology or at least an acknowledgement that I am not the only one out there with the problem would be much appreciated.

Answer 1

What Mac OS X version? What kind of computer? How much RAM?
I'm wondering if Acrobat may be running out of memory.