Abstract
This paper proposes a new fast algorithm for detecting human face and extracting the facial features. For this task, we have developed a flexible coordinate system and several support vector machines. The design of a face model for both detection and extraction is based on multi-resolution wavelet decomposition (MWD). Using a mean face, the MWD and a small number of feature points are applied for rough searching by estimating the modified cross correlation (MCC). More accurate results can be achieved by a serious of support vector machines (SVMs). Experimental results show that the proposed approach is fast and has a high detection rate even in cases when a face is embedded in a complicated background.
Original language | English |
---|---|
Pages (from-to) | 209-212 |
Number of pages | 4 |
Journal | Proceedings - International Conference on Pattern Recognition |
Volume | 16 |
Issue number | 4 |
Publication status | Published - 2002 |
ASJC Scopus subject areas
- Computer Vision and Pattern Recognition