The Coding Ecology: Image Coding via Competition among Experts

V. Michael Bove, Jr. and William Butera
MIT Media Laboratory
Proc. 1999 Picture Coding Symposium

We consider the image coding problem as a competitive ecology of specialists, each vying for the task of coding a portion of an image. Each specialist, or expert, derives its competitive advantage from its ability to concisely describe an underlying visual event ( e.g. shadows, motion, occluding objects). Experts locate candidate regions of support, characterize the activity within and submit their proposals in the form of a bid passed along to an auctioneer. In this paper, the metaphor of an auction informs the design of a predictive coder. We describe the designs for six experts, specify the cost function for comparing the bids, and report coding performance for several examples.