信用卡/借记卡中的数字细分

Question

I'm doing a project for which i need to segment credit card numbers from a credit card image.

I have gone through these post

http://rnd.azoft.com/optical-recognition-ios-application/

How to split credit card number digits into separate blocks for further recognition?

I do not want to use tesseract for bounding box or digit segmentation. Any method using opencv or any other image processing library is welcome.

Some of the input images are

Answer 1

If you refuse to use Tesseract, OpenCV do have some methods to perform digit segmentation. But the implementation is somehow a little bit more difficult than using Tesseract.

Basically, credit card digit extraction share the same process with license plate number extraction, and maybe the later one is more complex. So here are some tutorials you might need.

Number plate recognition code - Github

Licence plate detection with different backgrounds

Some method you might need:

Sobel filtering

Binarize using threshold function

Blob detection

Masking

SVM classifing

Answer 2

The characters are in printed form and are properly separated with space, hence you can easily tackle this problem using OpenCV

Follow below steps

1. Resize and grayscale the image
2. Apply morphological operation (to highlight light regions against a dark background)
3. Apply binarization and Otsu thresholding
4. Extract Contours (each Contour will correspond to a digit)