What steps will reproduce the problem?
1. Attached sample files can be OCRed using non .net wrapper
2. But cannot be OCRed using .Net wrapper; It gives all garbage
3.
What is the expected output? What do you see instead?
If the character are not broken, the .net wrapper works great. But the attached
images are out of dot matrix images.
If legacy Tesseract can OCR the sample images why not the attached one?
Also how can we update the "eng.Traineddata" file for .net wrapper. Especially,
if its possible to update the "eng.Traineddata" in legacy Tesseract.
What version of the product are you using? On what operating system?
tesseractdotnetwrapper_r590
Please provide any additional information below.
Original issue reported on code.google.com by sharmapr...@gmail.com on 27 Jun 2014 at 1:42
Original issue reported on code.google.com by
sharmapr...@gmail.com
on 27 Jun 2014 at 1:42Attachments: