Project Details
Abstract
Furthermore, by using the face plane we propose a new method that corrects inclinations of a facial attitude in a three-dimensional facial image. Such a method is very important as pre-processing in facial expression analysis or individual identification. Our method does not need to recognize facial parts such as eyes, a nose or a mouth in advance. In the experiment on correcting inclinations of a facial attitude, we compare the proposed coordinate system with an ordinary coordinate system obtained by using facial parts, and it is shown that our coordinate system is more effective than the ordinary coordinate system. In the experiment on recognizing a facial expression change, it is shown that our coordinate system is more effective too as pre-processing of a facial expression recognition system than the camera coordinate system.
Status | Finished |
---|---|
Effective start/end date | 2004/01/01 → 2005/12/31 |
Funding
- Japan Society for the Promotion of Science: ¥3,100,000.00
Keywords
- 顔平面
- 距離画像
- 分散最小基準
- 距離最小基準
- 表情差分マップ
- 表情分類
- 非接触3次元デジタイザ
- 顔座標系
- face plane
- distance image
- least variance criterion
- least distance criterion
- facial expression differential map
- facial expression classification
- non-contact 3-D digitizer
- facial coordinate system