The Case for Context-Aware Compression

The proliferation of pictures and videos in the Internet is imposing heavy demands on mobile data networks. This demand is expected to grow rapidly and a one-fit-all solution is unforeseeable. While researchers are approaching the problem from different directions, we identify a human-centric opportunity to reduce content size. Our intuition is that humans exhibit unequal interest towards different parts of a content, and parts that are less important may be traded off for price/performance benefits. For instance, a picture with the Statue of Liberty against a blue sky may be partitioned into two categories -- the semantically important statue, and the less important blue sky. When the need to minimize bandwidth/energy is acute, only the picture of the statue may be downloaded, along with a meta tag"background: blue sky". Once downloaded, an arbitrary "blue sky" may be suitably inserted behind the statue, reconstructing an approximation of the original picture. As long as the essence of the picture is retained from the human's perspective, such an approximation may be acceptable. This paper attempts to explore the scope and usefulness of this idea, and develop a broader research theme that we call context-aware compression.