Abstract | ||
---|---|---|
The Active Wavelet Network (AWN) [9] approach was recently proposedfor automatic face alignment, showing advantages over ActiveAppearance Models (AAM), such as more robustness against partialocclusions and illumination changes. In this paper, we (1) extendthe AWN method to a view-based approach, (2) verify the robustnessof our algorithm with respect to unseen views in a large datasetand (3)show that using only nine wavelets, our method yieldssimilar performance to state-of-the-art face alignment systems,with a significant enhancement in terms of speed. Afteroptimization, our system requires only 3ms per iteration on a1.6GHz Pentium IV. We show applications in face alignment forrecognition and real-time facial feature tracking underlarge posevariations. |
Year | DOI | Venue |
---|---|---|
2003 | 10.1109/AMFG.2003.1240846 | AMFG |
Keywords | Field | DocType |
face alignment forrecognition,method yieldssimilar performance,proposedfor automatic face alignment,active wavelet networks,illumination change,pentium iv,real-time view-based face alignment,activeappearance models,extendthe awn method,state-of-the-art face alignment system,view-based approach,active wavelet network,principal component analysis,learning artificial intelligence,face recognition,active appearance model,real time,feature extraction,wavelet transforms | Facial recognition system,Computer vision,Face hallucination,Pattern recognition,Three-dimensional face recognition,Computer science,Feature extraction,Robustness (computer science),Active appearance model,Artificial intelligence,Wavelet transform,Wavelet | Conference |
ISBN | Citations | PageRank |
0-7695-2010-3 | 16 | 1.15 |
References | Authors | |
14 | 3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Changbo Hu | 1 | 613 | 34.71 |
Rogério Feris | 2 | 1529 | 89.95 |
Matthew Turk | 3 | 3724 | 499.42 |