The testee is asked to make a specific utterance, and the image sequence containing the mouth motion is taken by a video camera. Lip motion is represented by the ratio (width vesus height of mouth) for each frame of the image sequence. The sequence of the opening ratio of the mouth is defined as "lip motion features".
1.Housing a cartoon image to dynamic frame grabber,DFM(Blay memory is 2GB)
2.Host Computer has one frame image(720x480) taken out.
3.Determination of the circumrectangle of the mouth
4.Extraction of lip motion features
5.Person identification by DP maching
Input image
Extraction of lip color area using HSL transform
Noise reduction using median filter
Expansion and shrink
Determination of the circumrectangle of the mouth
Examing existence of individual variation of lip motion features by equipment.
The utterance pattern is
In case of different utterance,the experimental results shows the 100% rate of another person dismissal and the 95% rate of acceptance in person. In another case of same utterance, the experimental results shows that this method realizes good person identification when length of utterance is not short.