PDF Renderer: allow to specify an alternate image or a custom resolution.

[phymbert]

Motivation

Input images passed to OCR are often pre-processed (higher resolution, grayed, etc...).
It can be useful to specify an alternate image or a lower resolution in renderer, especially for a searchable pdf export.

Proposed changes

1. Added TessResultRenderer::SetRenderingImage or TessResultRenderer::SetRenderingResolution methods allow to programmatically change image or resolution to render before adding image to the renderer
2. New rendering_dpi param allows to override the output resolution by scaling the source image
3. Added few pdfrenderer tests
4. Fix missing pdf.ttf font in the cmake install target

These changes might resolve #210 and #3798 features request.

Checks

• make check passed locally on ubuntu 23.10
• GitHub workflows passed

[zdenop]

cmake does not install a PDF font file. It was the old way, how to handle font in pdf. Now it is automatically included in library

[zdenop]

@jbreiden @jbreiden2 : Jeff can you have a look at this?

[phymbert]

cmake does not install a PDF font file. It was the old way, how to handle font in pdf. Now it is automatically included in library

Thanks, @zdenop, for the explanation. I was confused with the tessdata/Makefile.am, and I will remove it.
Let me submit pdfrendrerer test fixes, it failed on some platforms.

@jbreiden @jbreiden2, a better way to check pdf files generated than file maximum size is welcomed

[phymbert]

It looks there is a little interest, that happens :) Thanks all

[stweil]

Hi, it's not unusual that pull requests take some time before they are merged. That does not necessarily mean that there is little interest, but there is only a small number of people who contribute to pull requests by adding comments or testing them.

[phymbert]

No worries at all, I just saw it open on my to-do list for a while, so I preferred to close. Thanks for your feedback, I understand, reopened, no hurry.

[zdenop]

Since it extends the API functionality, it should be included in the 5.4.0 release.

[stweil]

I rebased this pull request and fixed a merge conflict.

[zdenop]

What about implementing this feature also to tesseract executable as a command line option?

[stweil]

Isn't that already possible with -c?

[zdenop]

Isn't that already possible with -c?

With -c I can set rendering_dpi. How can I set an image for SetRenderingImage?

[stweil]

Tesseract can create multi-page PDF files when it is called with a list of images. Ideally that should also work with alternate images.

[stweil]

Isn't that already possible with -c?

With -c I can set rendering_dpi. How can I set an image for SetRenderingImage?

Would it be possible to implement the desired features by only adding new Tesseract parameters – without any change of the C / C++ API?