How to make tesseract to recognize only numbers, when they are mixed with letters?

Question

I want to use tesseract to recognize only numbers. The problem is that I have mixture of numbers & letters and when I use SetVariable("tessedit_char_whitelist", "0123456789")
for every symbol tesseract returns wrong digit.

Can I set a threshold value so that tesseract omits the symbols with low resemblance?

NOTE: I set tesseract to recognize only digits so there is no confusion between O and 0.

Hi there, I am also using Tesseract with Java project and I am facing some issues, I have business cards images and I need to extract email addresses, the problem is that sometimes it makes confusion between numbers and letters, the email "j.schneiderheinze@dunskie.de" becomes "j.5chneiderheinze@dunskie.de", would you have and idea how to fix this? — Francisco Souza, Sep 16 '19 at 11:28

score 43 · Answer 1 · edited Nov 03 '20 at 12:41

Recognizing only numbers is actually answered on the tesseract FAQ page. See that page for more info, but if you have the version 3 package, the config files are already set up. You just specify on the commandline:

tesseract image.tif outputbase nobatch digits

As for the threshold value, I'm not sure which you mean. If your input is an unusual font, perhaps you might retrain with a sample of your input. An alternative is to change tesseract's pruning threshold. Both options are also mentioned in the FAQ.

score 15 · Answer 2 · edited Nov 04 '20 at 06:58

15

For tesseract 3, the command is simpler tesseract imagename outputbase digits according to the FAQ. But it doesn't work for me very well.

I turn to try different psm options and find -psm 6 works best for my case.

man tesseract for details.

edited Nov 04 '20 at 06:58

JMP

4,417
17
30
41

answered Apr 12 '13 at 03:24

michaelliu

1,667
2
13
13

The said digit-related info has been removed from the FAQ. – user598527 Jul 01 '23 at 05:50

score 14 · Answer 3 · edited Nov 12 '20 at 15:58

14

For tesseract 3, i try to create config file according FAQ.

BEFORE calling an Init function or put this in a text file called tessdata/configs/digits:

tessedit_char_whitelist 0123456789

then, it works by using the command: tesseract imagename outputbase digits

edited Nov 12 '20 at 15:58

Eric Lindsey

954
8
19

answered Jul 18 '14 at 08:44

user3852208

141
1
2

neoneye · Answer 4 · 2016-12-16T13:56:47.647

14

If one want to match 0-9

tesseract myimage.png stdout -c tessedit_char_whitelist=0123456789

Or if one almost wants to match 0-9, but with one or more different characters

tesseract myimage.png stdout -c tessedit_char_whitelist=01234ABCDE

edited Dec 16 '16 at 13:56

answered Dec 15 '16 at 12:41

neoneye

50,398
25
166
151

Blehi · Answer 5 · 2016-12-09T21:10:50.957

I made it a bit different (with tess-two). Maybe it will be useful for somebody.

So you need to initialize first the API.

TessBaseAPI baseApi = new TessBaseAPI();
baseApi.init(datapath, language, ocrEngineMode);

Then set the following variables

baseApi.setPageSegMode(TessBaseAPI.PageSegMode.PSM_SINGLE_LINE);
baseApi.setVariable(TessBaseAPI.VAR_CHAR_BLACKLIST, "!?@#$%&*()<>_-+=/:;'\"ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz");
baseApi.setVariable(TessBaseAPI.VAR_CHAR_WHITELIST, ".,0123456789");
baseApi.setVariable("classify_bln_numeric_mode", "1");

In this way the engine will check only the numbers.

score 5 · Answer 6 · answered Nov 23 '19 at 06:35

5

add "--psm 7 -c tessedit_char_whitelist=0123456789'" works for me when the image contain's only 1 line.

answered Nov 23 '19 at 06:35

Yerrick

141
2
3

score 4 · Answer 7 · answered Apr 24 '19 at 15:47

This feature is not supported in version 4. You can still use it via -c tessedit_char_whitelist=0123456789 with "--oem 0" which reverts to the old model.

There is a bounty to fix this issue.

Possible workarounds:

As stated by @amitdo

Using the --oem 0 option (the legacy engine will be used)
Retraining (fine tuning) #751 (comment)
Post-processing #751 (comment)

valentt · Answer 8 · 2014-08-26T17:34:31.743

3

You can instruct tesseract to use only digits, and if that is not accurate enough then best chance of getting better results is to go trough training process: http://www.resolveradiologic.com/blog/2013/01/15/training-tesseract/

edited Aug 26 '14 at 17:34

answered Nov 02 '13 at 00:45

valentt

857
8
19

1

I don't know why this answer is second when it's the least helpful and wrong. There is no **need** to go through training process again just to blacklist characters! – Arijoon Aug 15 '14 at 14:33
3

Blacklisting characters wasn't mentioned in my answer, but in some cases even if I had instructed tesseract to use numbers they weren't being recognized as such. In such cases training is only way forward. If you have gotten better results with some other process please share your experience. – valentt Aug 26 '14 at 17:36

score 3 · Answer 9 · answered Feb 25 '21 at 19:17

3

custom_oem=r'digits --oem 1 --psm 7 -c tessedit_char_whitelist=0123456789'

text = tess.image_to_string(croped,config=custom_oem)

I am using tesseract 4.1.1.

For better result you might want to consider Image processing techniques.

answered Feb 25 '21 at 19:17

Elie Eid

703
7
11

score -2 · Answer 10 · answered Aug 08 '15 at 21:47

-2

What I do is to recognize everything, and when I have the text, I take out all the characters except numbers

//This replaces all except numbers from 0 to 9
recognizedText = recognizedText.replaceAll("[^0-9]+", " ");

This works pretty well for me.

answered Aug 08 '15 at 21:47

algarrobo

304
1
12

5

You're missing the chance to improve quality and speed of OCR. If you restrict Tesseract to given set of characters, it will perform better. With your approach you're loosing valuable information. – Dmitry Zaytsev Sep 22 '15 at 13:14

How to make tesseract to recognize only numbers, when they are mixed with letters?

10 Answers10

Linked