Fails to recognize seemingly simple text

[Arjun Bk]

Hi, 

Attached here are two images that I cannot recognize using tesseract-ocr. I am calling a standalone .exe file (tesseract v4.0.0.20181030) in my python script to generate the pytesseract function. So far, I have not tried training my program. If anybody hints on that, I will try to learn it and apply. I wanted to know if there were some simple image enhancement solutions that I overlooked.

I have already tried to do a manual threshold, resize and invert of the image using OpenCV. Additionally, I wanted to specify that I only wish to detect the digits in each case.

Thanks.

[Lorenzo Bolzani]

Hi,

use psm 6 (or 7). Also try to crop to have a single line, if possible. Black text on white bg is better.

You should be able to isolate text in this way:

https://www.pyimagesearch.com/2017/07/17/credit-card-ocr-with-opencv-and-python/

Lorenzo

--
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 https://groups.google.com/group/tesseract-ocr.
To view this discussion on the web visit https://groups.google.com/d/msgid/tesseract-ocr/043e6ecb-36c1-44de-ad8b-b91734b24b59%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Arjun Bk]

Hi Lorenzo,

The link you shared was very helpful and your valuable suggestions helped me a lot. Now the image detection seems to work for at least 90% of my cases.

Just for information, I started with a top-hat transform followed by the Sobel operator and then applied the Otsu thresholding. Then I found the coordinates surrounding the largest two areas which gets me a box (with an allowance) that only covers the text I need (removing all other noises). Now the next important step was to determine if the image had to be inverted, for which I used a colour average. Then I finally used the suggested Page segmentation method (psm 6) to read my characters. Probably there could be better ways, but this works for the time being.

Thanks a lot. :)

-Arjun

On Thursday, 2 May 2019 19:31:39 UTC+2, Lorenzo Blz wrote:

Hi,

use psm 6 (or 7). Also try to crop to have a single line, if possible. Black text on white bg is better.

You should be able to isolate text in this way:

https://www.pyimagesearch.com/2017/07/17/credit-card-ocr-with-opencv-and-python/

Lorenzo

Il giorno gio 2 mag 2019 alle ore 16:15 Arjun Bk <abk...@gmail.com> ha scritto:

Hi, 

Attached here are two images that I cannot recognize using tesseract-ocr. I am calling a standalone .exe file (tesseract v4.0.0.20181030) in my python script to generate the pytesseract function. So far, I have not tried training my program. If anybody hints on that, I will try to learn it and apply. I wanted to know if there were some simple image enhancement solutions that I overlooked.

I have already tried to do a manual threshold, resize and invert of the image using OpenCV. Additionally, I wanted to specify that I only wish to detect the digits in each case.

Thanks.

--
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 tesser...@googlegroups.com.