How can i read a credit card number from an image in Matlab 2013a without using OCR

I have actually a list of credit cards in different format of image. I need to read only the credit card number, highlight it and save it in a new image fie. In attachement i have more images in different formats. please help me with the code. I write differrent codes but nothing helps with all and reading without OCR is difficult. I have 2013a install on my system. I will really appreciate the answers with code(if possible)

Thanks in advance

[Rik]

OCR stands for optical character recognition, so what you want to do is exactly OCR. Also, this is getting close to a project where it is easier to use it for a morally objectionable task than a legitimate one, so I will not provide practical advice. (of course I'm not suggesting you don't have a legitimate purpose, but on a public forum anyone will be able to use the advice)

[mohsin abbasi]

I agree. We know that we have machines that are doing these tasks very well. The submission of this question alongwith these pictures is to actually understand the complexity of this task and to develop the algorithm that can perform better than what we have. I have seen on Internet and realized that in matlab there is still need of algorithms for real world tasks.

Answer 1

[KALYAN ACHARJYA]

Have you tried with following links?

https://in.mathworks.com/matlabcentral/fileexchange/18169-optical-character-recognition-ocr

https://in.mathworks.com/matlabcentral/fileexchange/40499-optical-character-recognition-2-0-ocr-2-0

https://in.mathworks.com/matlabcentral/fileexchange/31322-optical-character-recognition-lower-case-and-space-included

[mohsin abbasi]

Unfortunately All these projects are using a simple image with text, written on a white background or a self made image that makes the text extraction quite simple. The real images as available on internet or taken from camera cannot be processed using them.

[KALYAN ACHARJYA]

One Suggestion: Recently I have done the OCR for bend images (Text line was not as straight). Such case you can do the the same OCR process for mutiple regions differently.