Facial Expression Recognition with PCA and LBP Features Extracting from Active Facial Patches

Facial expression recognition is an important part of Natural User Interface (NUI). Feature extraction is one important step which could contribute to fast and accurate expression recognition. In order to extract more effective features from the static images, this paper proposes an algorithm based on the combination of gray pixel value and Local Binary Patterns (LBP) features. Principal component analysis (PCA) is used to reduce dimensions of the features which are combined by the gray pixel value and Local Binary Patterns (LBP) features. All the features are extracted from the active facial patches. The active facial patches are these face regions which undergo a major change during different expressions. Softmax regression classifier is used to classify the six basic facial expressions, the experimental results on extended Cohn-Kanade (CK+) database gain an average recognition rate of 96.3% under leave-one-out cross validation method which validates every subject in the database.