Here�s how it works: whenver a SparkRebel user copies or uploads an image to the site, SparkRebel calls PicScout�s content identification service to recognize the image.  If it finds the image�s fingerprint in its database, it uses ImageIRC to determine who needs to be paid; then SparkRebel pays any royalty owed through Post Usage Billing.  PicScout ImageIRC includes Getty�s own images; it is the largest stock image agency in the world.  (Getty Images itself was sold just last month to Carlyle Group, the private equity giant, for over US $3 Billion.)  In total, ImageIRC includes data on over 80 million images from more than 200 licensors, which can opt in to the arrangement with SparkRebel (and presumably similar deals in the future).

Where does the money come from, though? The person using the image must have to agree to pay for it. Do you suppose this is just another way for Getty to make more money? How much money will actually be passed along to the contributor? How much money does SparkRebel then PicScout then Getty take? It sounds like a good idea but if Getty's name is on it, rest assured the contributor will get the short end of the stick. 

So they use image recognition to find out the copyright owner?

Most of the times image recognition just doesn't work.
And when it works, what if they find the picture on another photo sharing site? Will they pay someone else for stealing our work?