ICA and Gabor representation for facial expression recognition

Two hybrid systems for classifying seven categories of human facial expression are proposed. The £rst system combines independent component analysis (ICA) and support vector machines (SVMs). The original face image database is decomposed into linear combinations of several basis images, where the corresponding coef£cients of these combinations are fed up into SVMs instead of an original feature vector comprised of grayscale image pixel values. The classi£cation accuracy of this system is compared against that of baseline techniques that combine ICA with either two-class cosine similarity classi£ers or two-class maximum correlation classi£ers, when we classify facial expressions into these seven classes. We found that, ICA decomposition combined with SVMs outperforms the aforementioned baseline classi£ers. The second system proposed operates in two steps: £rst, a set of Gabor wavelets (GWs) is applied to the original face image database and, second, the new features obtained are...

• Type: Conference
• Year: 2003
• Where: ICIP
• Authors: Ioan Buciu, Constantine Kotropoulos, Ioannis Pitas
• Keywords: Cosine Similarity Classi£ers, Facial Expression, ICIP 2003, Image Processing, Maximum Correlation Classi£er