While lately the content of slashdot has really been lame, there was a nice article yesterday about a algorithm for content aware resizing. The algorithm presented at SIGGRAPH 2007 by Shai Avidan and Ariel Shamir is able to resize images by selectively removing parts of the image with the least information. This is useful to make a picture fit within a frame with a determined ratio, and more graceful than distorting the picture, clipping out the border, or adding a black border.
This method for rescaling images does not preserve the image, basically the algorithm cuts out regions that do not contain meaningful data (the actual metric for this can be changed), the movie actually shows how to give hints to the algorithm to perform actual retouching of pictures, removing persons. Of course, all this was doable before by a person using some digital retouching software, but the possibility of doing it in an automated way could have quite some implications.