Vertical code detection with Tesseract OCR

Viewed 31

I'm trying to use tesseract for a text recognition of containers ILU codes. I have issue with vertical code orientation. Before using the photo with tesseract, I edited it like this one: image.png

I set this configuration parameters:

  • tessedit_char_whitelist - 0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ
  • psm – 11 (I also try 5)
  • load_system_dawg – 0
  • load_freq_dawg – 0
  • enable_new_segsearch – 1
  • language_model_penalty_non_freq_dict_word – 1
  • language_model_penalty_non_dict_word - 1

I'm using tesseract 5.2.0-22-g0daf1 in a C++ application and I use result iterator the same way as in the api examples (https://github.com/tesseract-ocr/tessdoc/blob/main/APIExample.md). This issue was already mentioned in this post: detection of vertical texts (container BIC codes) with tesseract OCR fails. I tried to set the same parameters, but haven't received any satisfactory result. Is there any other parameter that should be set?

0 Answers
Related