Unnatural image block in the left box represents one state while natural one in the right box another state. Difference of the image states is used for data hiding and, as a consequence, data compression. Courtesy of Korea IT Times

The conventional wisdom was that already-compressed files cannot be compressed again.

However, a new exquisite image compression technology that had been viewed as unobtainable by many has been successfully developed by a Korean image processing guru for the first time in the world. Hyoung Joong Kim, professor at Graduate School of Information Security, Korea Univ. has succeeded in compressing image files in the JPEG format (a standard method of lossy plus lossless compression for digital photography).

Generally speaking, the probability of every symbol is put to use at the final stage of bit-rate reduction or data compression. Since compression algorithms have fully exploited the probability of every symbol, making attempts to compress them again by reusing the probability of each symbol is impossible. Therefore, Kim has come up with a trailblazing data compression method: hiding information in compressed files using a reversible data hiding technique. His compression algorithm embeds message into the compressed bit stream, and hidden information can be extracted and compared with the uncompressed image states.

Read more.

Source: Korea IT Times