Tool automates photomontage edits

Researchers from the University of Washington and Microsoft Research have devised a method of automatically and seamlessly combining features of multiple photographs, including photographs that depict movement.

The method, dubbed Interactive Digital Photomontage, automatically finds seams between parts of two or more source photographs after a user provides rough guidance.

The software allow users to, for instance, make a composite photograph from several pictures of a group of people, according to the researchers.

The system also allows users to make infinite depth-of-field images of microscopic subjects. Macro lenses that take extreme close-ups have short depths of field, making it difficult to get the whole picture in focus at once. The software can stitch together images captured at different focal lengths to form a sharp composite, according to the researchers.

The method uses a graph-cut optimization technique to choose good seams for segmenting source images, and a gradient-domain fusion technique to smooth color differences along the seams once the composite is put together.

It also includes a suite of software tools that allow the user to specify high-level image objectives for particular areas or globally using a paint-style interface. To make a composite photograph from several pictures of a group, for instance, the user selects the best depiction of each person by drawing a line over it.

The method can be used practically now, according to the researchers. The researchers are scheduled to present the work at the Association of Computing Machinery (ACM) Siggraph 2004 conference in Los Angeles, August 10 to 12.