Tesserect : Recognized Text From Image come up with very Poor Quality about tess-two HOT 12 CLOSED

You can improve quality by set OEM mode like this (make sure you copying all train data files into tessdata folder):
baseApi.init(tessDataPath, "eng", TessBaseAPI.OEM_TESSERACT_CUBE_COMBINED);

And there are many preprocessing image methods, grayscale is just one of them.Take a look at this http://mitako.eu/OcrTips and http://www.cra.org/Activities/craw_archive/dmp/awards/2008/Elmore/melmore_dmp.pdf

from tess-two.

Hi.. thank you for reply
.I m already tryied what you said. but when i am set TessBaseAPI.OEM_TESSERACT_CUBE_COMBINED mode it raise following error...
Fatal signal 11 (SIGSEGV) at 0x00000000 (code=1), thread 5097 (roid.ocr.simple)
what is wrong in it..?

from tess-two.

As I said, make sure you copying all train data files with your language into tessdata folder (for example, with the english language, there are many files eng.*) and it should be placed in phone or emulator sdcard or internal storage, download from here https://code.google.com/p/tesseract-ocr/downloads/list

from tess-two.

hey dactranit..
I alreadt done.
In my project directory having folder with tessdata and in that eng.traindata file exixt

from tess-two.

anyother train data files requir..?

from tess-two.

Not only eng.traindata, I said all eng.* files, so they are eng.cube.bigrams, eng.cube.fold ... (you should copy ALL files in this package https://tesseract-ocr.googlecode.com/files/tesseract-ocr-3.02.eng.tar.gz) please PUT THEM INTO PHONE OR EMULATOR 's STORAGE OR SDCARD, INSTEAD OF PROJECT DIRECTORY. If still not work, please provide your code and a screenshot of file explorer in your device that contain these files.

from tess-two.

Hi...I am done as u said and it work fine but i am not get the best accuracy as it commented...
i am get same result as the "OEM_TESSERACT_ONLY" gives me

from tess-two.

• Make sure your image should be black and white image (or grayscale ) because tesseract work better with it. You can google how to convert image into grayscale
• PLEASE read the document (https://code.google.com/p/tesseract-ocr/w/list) carefully before asking, there are a lot of advice and guides. If you want to improve the quality, you should have a pre-processing image before put into tesseract, read this: https://code.google.com/p/tesseract-ocr/wiki/ImproveQuality

from tess-two.

Hii ..Thank you for reply..
One more Question i would like to ask you...
What is the Advantage if we disable dictionary..?

Waiting for your reply..

from tess-two.

Read this https://groups.google.com/forum/#!topic/tesseract-ocr/4XHtbqaKAmQ
By using dictionary, tesseract indicates whether to correct the word or not, but it doesn't make sure work as aspect. When disable dictionary, you can improve the performance of OCR process, make it faster, but the result quality may be poor.

from tess-two.

Hey which PSM mode is best to use..?

from tess-two.

This is a question and not really an issue.

Also see:
https://github.com/renard314/textfairy
https://github.com/artetxem/mitzuli

from tess-two.