MACHINES could soon be able to read people鈥檚 facial expressions, no matter how subtle they are.
So-called automatic expression analysis is difficult because people鈥檚 faces vary so much. Existing software can effectively recognise only extreme expressions, which is not particularly helpful in real-life situations.
Now Sungsoo Park and Daijin Kim at in South Korea have developed a system that recognises even slight facial movements. They programmed software to identify 27 facial feature points, such as the position of eight points around the mouth and three around the nostrils, and to track how each point moved in a sequence of images for each facial expression. The system then exaggerates the movements in each sequence, generating extreme facial expressions which existing software can identify.
Advertisement
In tests in which 80 subtle expressions of neutral, happy or angry feelings were exaggerated in this way, the system judged them correctly 88 per cent of the time (Pattern Recognition Letters, ).
The team has used the technique to allow a robot with a human-like face to read and mimic a person鈥檚 subtle expressions. They also hope to install the software on digital cameras and camera phones, says Kim, so the devices can do things like automatically take a picture when the subject smiles.
鈥淒igital cameras with the software could take a picture automatically when someone smiles鈥�
