Hello,
I want to start a project that relies upon dice face value recognition. The idea is using a kinect pointed at a glass surface and detecting various colored dice.
I need sugestions on how to do recognize the values/patterns of the dice.
My own suggestion is detecting a colored blob (which is roughly a square by nature). Inside the blob check how many pixels are a diferente color and transpose that ratio (blobColor/otherColor) to a numerical value which hopefully will give me the correct reading.
How do you guys think?