Recently I read an article about two new tools that are being developed to help digital photographers remove items from their images. Suppose you had taken picture of a great small town main street, but when saw the image you realized that three parked cars obscured the view and just looked messy. If only you could remove them! However, if you did, you would be left with a void.
Apparently, you are not alone. At least two teams are trying to solve this problem. One is developing a scene-completion algorithm that searches for a patch for your picture. It quickly scans thought a database of millions of images on Flickr, looking for images of the same subject, taken from the same position, with the light falling from the same direction. The algorithm narrows down all the choices first to 200 and finally to 20 or so. Ultimately, user are given a choice and select the one that looks best to them.
Another team is trying to create a library of clip art, taken from the Label Me library of images. This library could supply patches for missing picture parts. When these tools get perfected, they may allow photographers to save pictures that otherwise would have to be scrapped.