Poker card recognition with computer vision methods

In this paper, the face of playing cards is used as the identification target carrier. The card picture to be recognized is a 24-bit true-color image taken by a mobile phone, and then the image is preprocessed to a 256-grayscale picture, and then further converted to only 0 and 1 binary image. Finally, the contour shape matching is carried out according to the templates of hearts, spades, diamonds, clubs, and Ace-King obtained in advance. The contour shape matching is based on the Hu invariant moment. The nature of the Hu invariant moment shows that the rotation of playing cards does not affect it. Playing card recognition can assist visually impaired people to play poker games so that some people can also feel the fun of poker games and enjoy it. The playing card recognition can also realize the digital storage of cards, which can make the game fairer.