High Quality 3D Facial Images From Surveillance Video

GE Global Research will develop the Face Enhancer system which will consist of a near real-time PC-based solution taking surveillance video as input. Automatic face detection will scan the video for potential subjects. A 3D face model will be aligned to each detected individual in the video sequence. 3D surface super-resolution techniques will use the aligned 3D face model and combine the video frames to synthesize a facial image which has up to 50% higher resolution than a single frame. The resulting high-resolution 3D facial surface will be rotated and rendered so that the subject can be viewed from an arbitrary pose. ca/ncf