[
https://issues.apache.org/jira/browse/TIKA-2749?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16810172#comment-16810172
]
Ross Johnson commented on TIKA-2749:
------------------------------------
OCRing the inlined images directly can be tricky, in my experience. Here are a
couple classes of problematic scenarios that I've come across:
- Sometimes images have a funky transform applied, e.g. the actual page scan is
mirrored but flipped to look right in the PDF.
- Some fancy PDF generators / scanners use multiple overlapping images, perhaps
utilizing image masking, to reduce file size. E.g. there may be a background
image with the color components along with a foreground grayscale or
1-bit-per-pixel image of the black content. I've also seen the foreground text
split up into multiple images overlaying different parts of the background
image.
In either situation, you could probably detect the "not a normal scan"
condition and kick it into "render page" mode.
> OCR on PDFs should "just work" out of the box
> ---------------------------------------------
>
> Key: TIKA-2749
> URL: https://issues.apache.org/jira/browse/TIKA-2749
> Project: Tika
> Issue Type: Task
> Reporter: Tim Allison
> Priority: Major
>
> There are now two different ways (with various parameters) to trigger OCR on
> inline images within PDFs. The user has to 1) understand that these are
> available and then 2) elect to turn one of those on.
> I think we should make OCR'ing on PDFs "just work" perhaps with a hybrid
> strategy between the 2 options. Users should still be allowed to configure
> as they wish, of course.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
