University of Washington researchers have devised a method to capture thousands of digital photos of a person’s face and then create an animation of the face within seconds.
The software has the ability to change the facial expression images of a person from a smiling face to a frown. It can also compose the face and make it to appear as it ages over time.
UW researcher in computer science and engineering, Ira Kemelmacher-Shlizerman, is to present the study at the Special Interest Group on Graphics and Interactive Techniques meeting in Vancouver.
The project builds on an earlier research, which stitched photos of buildings into a three-dimensional scene. A person’s face changes and ages over time, which presents more challenges. Developments in face detection technology have improved the feasibility of the project. Picasa, iPhoto, Windows Live Photo Gallery and Facebook have incorporated face-recognition tools.
The project led to a video of a girl as she aged from birth till she attained 20 years of age. Older photos were scanned, tagged with the girl’s name and dates were then added. A less-than-a minute movie of the photos was made, which showed the aging process over 20 years. The Face Movie feature in the Picasa photo tool is one of the versions of the tool being developed.
This research focuses on unstructured photos that are taken under varying conditions, while other projects focused on photos that were taken under controlled conditions.
Research on utilizing personal photos to construct three-dimensional models of a face is to be presented at the International Conference on Computer Vision. The three-dimensional models have the potential to be used for recognizing faces and for transmitting a person’s face during video conferencing.