Image Inversion vs Border Creation [Docs Question]

[Philip L]

At https://github.com/tesseract-ocr/tessdoc/blob/main/ImproveQuality.md#binarisation, the Tesseract docs say:

While tesseract version 3.05 (and older) handle inverted image (dark background and light text) without problem, for 4.x version use dark text on light background

and

If you OCR just text area without any border, tesseract could have problems with it. See for some details in tesseract user forum#427 . You can easy add small border (e.g. 10 pt) with ImageMagick®:

convert 427-1.jpg -bordercolor White -border 10x10 427-1b.jpg

I'm a little puzzled about two things:

1. If we're using a light background, won't "adding a white border" typcially just mean making a larger image with the target text making up less of its area (because the border will match the color of the background)? Is that the intended interpretation of this -- to avoid text that directly touches the boundaries of the image?

2. The inversion advice talks about Tesseract 3 and 4. Does Tesseract 5 maintain the "dark text on light background" preference of 4?

p.s. Tried to post a message once before and it didn't show up for some reason. Giving it one more shot; sorry if this doubleposts.

[Ger Hobbelt]

Your #1 is correct.

Tesseract4 behaviour here applies to tesseract4 without change. (In other words: tesseract4 is the odd one out)

--
You received this message because you are subscribed to the Google Groups "tesseract-ocr" group.
To unsubscribe from this group and stop receiving emails from it, send an email to tesseract-oc...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/tesseract-ocr/4fc6a9e9-b674-479b-930b-e955f23204d1n%40googlegroups.com.