0

I want to make a printable scannable code, something like snapcode or Spotify code. The code will have a grid with three rows and five columns of coloured dots. Each coloured dot will point to a number or character so that a hash will be formed.

I would like to start with a way to find the coloured pattern of the scanned code. So that each colour and position of the dots will be known, to construct the hash. The code will be scanned with smartphones or tablets.

In my search, I saw Hough transformation, Image segmentation/moment analysis/templating, Surf, Canny edge detection and more. Lots of ways. But I'm wondering, what the best way is to (always correctly) recognize the coloured pattern?

Edit, added images:

smart code with logo smart code still from camera

Jeroen Steen
  • 531
  • 8
  • 22
  • What's wrong with a QR code? – Mark Setchell Feb 21 '18 at 09:18
  • @Mark A QR-style code (like Snapcode) can also be something. But a QR-style code and QR-code is different, right? I would like to make it more unique then a regular QR-code (three big squares in the corner), even though I know it can be customized with a logo in the center for instance. – Jeroen Steen Feb 21 '18 at 09:37
  • Maybe you could add an image so folks can get an idea of the likely problems with detecting/decoding it? – Mark Setchell Feb 21 '18 at 09:42
  • 1
    style > function is rarely a good choice. you could use something like SURF to detect the diamond, then read the red and blue dots using a mask. without defining what is good you cannot ask for what is best. try several approaches and see for your self. things like that are very application specific. https://docs.opencv.org/2.4/doc/tutorials/features2d/feature_homography/feature_homography.html – Piglet Feb 21 '18 at 10:57

0 Answers0