A player will be considered:

• seated, if his chair is occupied

• active: if he is posting blinds, i.e not sitting out

• playing, if he holds cards; i.e either face-cards or cardbacks.

OpenHoldem will scrape the above information in this order and much depends on getting this information right. For example if a player is not even seated, then OpenHoldem will be smart and avoid the costly scraping of cards and especially names. Fortunately deciding if a player is seated or active can usually be done with an inexpensive and reliable hash-transformation, which usually requires only the colour of a single pixel.

Scraping cards works in a similar way: It requires:

• cardbacks: again this information can usually be scraped with a cheap and reliable colour-transform

• “nocard”-regions. These regions were introduced for two reasons: if a player holds no cards we can avoid the somewhat costly scraping for card-faces which usually requires fuzzy fonts. As an additional benefit we get improved error-detection in debug-mode, because a player either shows cardbacks, cardfaces or no-card.

• cardfaces (or alternatively ranks plus suits): this information can sometimes be done with hashes or images, if the cards are always displayed the same way: but often the display will differ a little bit and you will need text-scraping (fuzzy fonts).

In practice it looks like this:

Image	Situation	Parameters
	Not seated	pXseated == false Rest does not matter
	Seated, but sitting out	pXseated == true pXactive == false
	Seated and active, but not playing	pXseated == true pXactive == true pXcardback == false pYcardfaceYnocard == true
	Seated, active and playing (villain)	pXseated == true pXactive == true pXcardback == true
	Seated, active and playing (usually hero, at some casinos villain at showdown)	pXseated == true pXactive == true pXcardback == false pXcardfaceYnocard == false