1

The images have the following features (as attachedNumbers squeezed, with noisy drawing):

  • The written numbers are squeezed in a square-shaped table on the image
  • They don't share the same length
  • Some numbers even have noisy drawing around them

What I need to do is to extract these numbers into readable format, like excel and upload to our ERP system. I googled but did not find suitable solution for this one

Aquaman
  • 11
  • 1
  • 1
    Give this [article](https://stackoverflow.com/questions/9413216/simple-digit-recognition-ocr-in-opencv-python) a read. Here the accepted answer makes use of custom image dataset for detecting numbers. You can do the same thing for your work, but use the [MNIST Digits dataset](http://yann.lecun.com/exdb/mnist/), since your image has handwritten data. – Argon Jun 22 '19 at 04:07
  • Thanks very much for your reply, Argon. I am reading the article you mentioned now and see whether my problem can be solved. – Aquaman Jun 25 '19 at 06:33
  • @Argon Thanks Argon. It seems that I got a problem that it is not easy to locate the table precisely (the image attached is only part of raw image). I tried to use match template in opencv but failed to locate it correctly. Any suggestion for me to going forward? – Aquaman Jul 08 '19 at 12:50

0 Answers0