Facial Region Segmentation Based Emotion Recognition Using K-Nearest Neighbors

As part of human emotion recognition, a As part of human emotion recognition, a facial expression recognition method is proposed where segmentation of facial regions are done manually in a unique yet effective way by analyzing many human faces and the position of the right eye, left eye, nose and mouth in those faces. Feature extraction from the segmented parts are done using 2D Gabor filter, redundant features are eliminated using downsampling of the extracted features and finally, classification of the expressions are done using K-Nearest Neighbors (KNN) classification technique. To evaluate the performance of the proposed method CK+, RaFD and KDEF datasets are used. For the CK+, RaFD and KDEF dataset the recognition rates are 99.75%, 94.60% and 86.02% respectively which indicates the effectiveness of the proposed method.