You are viewing limited content. For full access, please sign in.

Question

Question

OCR multi-directional text from Maps

asked on January 21, 2014

What is the best way to OCR multi-directional text from Maps?

0 0

Replies

replied on January 28, 2014

It seems like you'd want to OCR it in two orientations and filter the output (maybe through a dictionary) to get rid of mis-recognized characters, and then save that text to the page.  Unfortunately, I can't think of a way to automate this using Laserfiche products.  You might be able to put something together with the SDK.  You would have to save the PDF as an image first (otherwise there's no control over page orientation), and I think our OCR operations overwrite text and don't give you a hook to validate the text, so it becomes a somewhat circuitous sequence of operations.

1 0
replied on January 22, 2014

Depends on the text and the map. If you attach an example, we can try to give suggestions. smiley

0 0
replied on January 28, 2014

Hi Ege,

 

I attached a document last week.  Please let me know if you have any additional information.

 

Thank you,

Dan

0 0
replied on January 22, 2014

Please see the attached sample document.

Test Map.pdf (3.96 MB)
0 0
replied on February 3, 2014

It sounds like Quick Fields OmniPage Zone OCR might do what you need: it allows you to specify OCR settings for a portion of the image, including rotation for that section only. Note that the rotation angles are limited to 90 degree increments, and I you have some diagonal text, so you may not be able to get all the text OCR'd this way.

 

That said, Quick Fields has a lot of processing capabilities; perhaps try using local enhancements for each zone OCR you want to perform, such as crop and then de-skew before running the OCR.

 

Since you have to set up a decent number of zones for this map, Quick Fields may not be a time-saver if each map is drastically different.

You are not allowed to follow up in this post.

Sign in to reply to this post.