Facial expression analysis by kernel figenspace method based on class features (kemc) using non-linear basis for separation of expression-classes

Human Emotion Recognition and Classification from Digital Colour Images Using Fuzzy and PCA Approach

“…In this, gabor features are extracted using gabor filter banks and compressed by two stage PCA method. Kernel Eigen Space method based on class features for expression analysis is explained in [7].…”

Texture and shape information fusion for facial expression and facial action unit recognition

“…We examined the FAUs 1,2,4,5,6,7,9,10,12,15,16,17,20,23,24,25 and 26, as proposed in the facial expression recognition rules in [5] (17 FAUs in total).…”

“…For the kth FAU recognition, the database is clustered into two different classes {V (1) k , V (2) k } each one representing one possible kth FAU state (presence or absence). The grid deformation feature vector g j ∈ R Q is used as an input to 17 two-class SVM systems, each one detecting a specific FAU (the FAU set includes FAUs 1,2,4,5,6,7,9,10,12,15,16,17,20,23,24,25 and 26). Each SVM system, uses the Candide node geometrical displacements to decide whether a specific FAU is activated for the test grid under examination or not.…”

“…The most frequently used texture features are Gabor filter output [8][9][10], pixel intensities [11][12][13][14][15][16], Discrete Cosine Transform (DCT) features [15] and skin color information [17][18][19]. Accordingly, feature extraction methods based on Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA) [13,20] have been used in order to enhance the performance of texture information.…”