Wave goodbye to bad photographs!

A tool is being developed that can identify similar images to your photo from an online library, so that elements can be added in to your picture in place of objects you wish to delete.

This new development comes from researchers at the illustrious Carnegie Mellon University in Pittsburgh. Here teams have been working on a tool that searches millions of images on sites such as Flickr to identify similar elements to the photo you are editing, and gleans appropriate parts of it to replace unwanted elements of your original photo. This means that if there’s a huge building spoiling an otherwise idyllic view, then the building can be removed and an image similar to the surrounding view added in. The researches have created an algorithm that uses light sources, camera position and composition in a photo to detect a similar match to the photo. The tool searches for appropriate pictures amongst the 2.3 million images on Flickr and eliminates over 99.9% of them straight away, and only chooses the closest 200 for closer scrutiny. These 200 images are then analysed for elements such as hillsides or seascapes of the appropriate size and colour of the deleted object. From these 200 the twenty best images are then cropped and their useful parts added to the scene. The head of the team carrying out the research has said that the hardships lie in creating images which are highly credible, whilst still making them available in an easy-to-use library format.

The algorithm has recently been revealed at the Siggraph computer graphics conference in San Diego, and a research paper published describing the development of the tool in detail. Now photography fans will have to wait to see whether the program is developed for public consumption. If it does it could mean the end to ruined photos and unwanted people popping up in your holiday snaps.

Source:

BBC News