I'm having problems to OCR images that are not clear. Images with a dirty background making hard to distinguish the text.
I would like to apply filters to make the text more clear. I'm opened to suggestions. My first idea is to try to filter all pixels that are black. How can I do that in ImageEn?
As far as I know... there are no ImageEn procedures that will do this satisfactorily. I have tried many times to clean my scans (remove black dots and extraneous blotches) using various procedures without much success.
Having said that, Nigel has expressed that he would like to look into this, but without some sort of test images he will have nothing to work with.
If you really need this feature then I suggest you either get permission from your client to send an image or two to the ImageEn staff or create some similar images with your scanner with the same bit-depth. You do not have to post the images. Just send them directly to Nigel via email.
A RemoveIsolatedPixels procedure that will work with black & white, 8-bit or 24-bit images would be a great addition to ImageEn that most everyone who uses ImageEn scanner functions would value highly.
If we work together maybe we can develop some method(s) that work to clean images with the typical artifacts that you get when using scanners. If not... you are stuck with the images you scan.
Search
Filter to improve OCR
Posted - Sep 22 2014 : 13:32:09
