Training tesseract to recognize text from images

[tesseract-ocr]

Training tesseract to recognize text from images

[chrys87]

Hi,
for me it helps often to upscale a image.
in ocrdesktop i resize the image by factor 3 (Bicubic). This brings quite good results.
Cheers

[Wikinaut]

In case that you use screenshots, please notice, that screenshots usually have 72dpi, which is not sufficient for Tesseract. I admit, that this is not so well known, however, it is mentioned since a long time here https://github.com/tesseract-ocr/tesseract/wiki/FAQ#is-there-a-minimum-text-size-it-wont-read-screen-text

Is there a Minimum Text Size? (It won't read screen text!)

There is a minimum text size for reasonable accuracy. You have to consider resolution as well as point size. Accuracy drops off below 10pt x 300dpi, rapidly below 8pt x 300dpi. A quick check is to count the pixels of the x-height of your characters. (X-height is the height of the lower case x.) At 10pt x 300dpi x-heights are typically about 20 pixels, although this can vary dramatically from font to font. Below an x-height of 10 pixels, you have very little chance of accurate results, and below about 8 pixels, most of the text will be "noise removed".

Here's another useful page:
https://github.com/tesseract-ocr/tesseract/wiki/ImproveQuality

I recommend:

• remember never to use JPEG (because this is a lossy compresison format) when saving images with text (use: PNG, or use TIF, or GIF as the last choice)
• resize your screenshots or images to at least 300dpi, or 400dpi, or upscale by 400%:

convert -resize 400% ...

• when you start form a pdf then read http://bertanguven.com/faster-conversions-from-pdf-to-pngjpeg-imagemagick-vs-ghostscript/

density=400
gs -dNOPAUSE -sDEVICE=png16m -sOutputFile=$image -r$density -q $file -c quit

[Posting updated with information which came in later]

[amitdo]

@Wikinaut, why do you link to the old wiki at googlecode instead to the new one at github?
https://github.com/tesseract-ocr/tesseract/wiki/FAQ#is-there-a-minimum-text-size-it-wont-read-screen-text
Here is another useful wiki page.
https://github.com/tesseract-ocr/tesseract/wiki/ImproveQuality

[Wikinaut updated his comment.]

[Wikinaut]

I updated my posting above accordingly. Thanks

[zdenop]

Please use tesseract user forum for support[1]. Also do not forget to search forum for you topic before asking for help.
[1] https://github.com/tesseract-ocr/tesseract/wiki/FAQ#rules-and-advice

[suraj990]

open
raise ValueError("bad mode %r" % mode)
ValueError: bad mode 'sign.png'

what to do i am helpless