Read the Original Article at http://www.informationweek.com/news/showArticle.jhtml?articleID=227400397
Video content is created in a variety of aspect ratios, a ratio of image width to height. HDTV content usually has a 16:9 aspect ratio; content created for television comes in a 4:3 aspect ratio. A 640 by 480 pixel computer screen is also 4:3, or 1.33:1. A variety of other aspect ratios (1.85:1, 2.39:1, 2.66:1) are used in the film industry.
This becomes a problem when images are displayed on screens with a different aspect ratios. The iPhone 3GS screen, for example, is 480 by 320, or 1.5:1. Images may be resized to make them fit, but resizing images for a different aspect ratio leads to distortion -- the image is either stretched or squeezed as the aspect ratio is increased or decreased.
The alternative, image scaling, isn't ideal either. Scaled images maintain the same ratio of image width to height but the result won't fill a screen of a different aspect ratio without image loss (cropping) or black bars -- something most people who view DVDs of motion pictures on their TVs have seen.
But researchers from Google and Georgia Tech have come up with a new way to change the aspect ratios of images without noticeable distortion or image loss. They call the process "retargeting" to distinguish it from resizing.
In a post on the Google research blog, Georgia Tech PhD candidate Matthias Grundmann and Google research scientist Vivek Kwatra describe the work they have done with Georgia Tech interactive computing professor Irfan Essa and Google researcher Mei Han.
"We have developed an algorithm that resizes (or retargets) videos to fit the form factor of a given device without cropping, stretching or letterboxing," state Grundmann and Kwatra in their post. "Our approach uses all of the screen’s precious pixels, while striving to deliver as much video-content of the original as possible. The result is a video that adapts to your needs, so you don’t have to adapt to the video."
The algorithm uses a technique called discontinuous seam-carving to alter sequences of video frames by leaving salient details like faces intact while compressing or stretching non-salient areas like backgrounds. The result is an image that appears to be made for the screen size upon which it is displayed.
An example of how retargeted video looks can be seen in the YouTube video below.