poor results on relatively straightforward image

[hj]

see attached image. have tried various things, including this config:

tessedit_char_whitelist 0123456789:.

load_system_dawg F

load_freq_dawg F

tessedit_write_images T 

tessedit_create_boxfile T

and playing around with the source image in various ways to get it in its attached form. also run it with -psm 7 (and have tried 8) to no avail.

currently i get: "30:59" which is significantly better than what i was obtaining previously. 

based on the previous two posts and one i found on stackoverflow,  it seems like tesseract has difficulty classifying "digital" font characters. 

any suggested preprocessing that would improve performance here?

[Dmitri Silaev]

Show the source image. Show what you have done to get the binarized version.

Best regards,
Dmitri Silaev
www.CustomOCR.com

--
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 post to this group, send email to tesser...@googlegroups.com.
Visit this group at http://groups.google.com/group/tesseract-ocr.
To view this discussion on the web visit https://groups.google.com/d/msgid/tesseract-ocr/30dcc5f5-d7f3-4916-8dc4-cc8ad08b3d6b%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.