Abstract | ||
---|---|---|
In this paper, we introduce an advanced age determination technique that combines a feature set derived from an image of the face using multi-factored Principal Components Analysis (PCA) on the shape of the face and its features and the skin of the face to produce a 30 × 1 linear encoding of the face. The linearly encoded features are combined with Spectral Regression (SR) to improve performance of age determination over the current best techniques. The technique of SR is used to further reduce the dimensionality of the face encoding such that inter-class distances are minimized while maximizing intra-class distances. The SR feature vector is used to classify a face into one-of-two age groups (age recognition). An age-determination function is constructed for each age group in accordance to physiological growth periods for humans - pre-adult (youth) and adult. Compared to published results, this method yields the highest accuracy rates in overall mean-absolute error (MAE), mean-absolute error per decade of life (MAE/D), and cumulative match score. |
Year | DOI | Venue |
---|---|---|
2010 | 10.1109/CVPRW.2010.5544612 | CVPR Workshops |
Field | DocType | Volume |
Facial recognition system,Feature vector,Pattern recognition,Regression analysis,Computer science,Curse of dimensionality,Feature extraction,Artificial intelligence,Statistics,Principal component analysis,Spectral regression,Encoding (memory) | Conference | 2010 |
Issue | ISSN | Citations |
1 | 2160-7508 | 7 |
PageRank | References | Authors |
0.47 | 5 | 4 |
Name | Order | Citations | PageRank |
---|---|---|---|
Khoa Luu | 1 | 200 | 26.05 |
Tien Dai Bui | 2 | 33 | 4.40 |
Ching Y. Suen | 3 | 7569 | 1127.54 |
Karl Ricanek | 4 | 165 | 18.65 |