Abstract

In recent year, Emotion recognition can be done through different modalities, such as speech,
facial expression, body gesture etc. Emotion recognition through facial expression has attracted a lot of
interest. Generally the emotions are classified as anger, disgust, fear, happiness, sadness, and surprise.
The accurate prediction of facial expression is difficult due to the illumination, pose facial occlusions and
different shape. To solve this problem the previous system introduced salient facial patches based facial
expression recognition. However it has lower true positive rate. In order to improve the face recognition
accuracy, the proposed system introduced Improved Artificial Bee Colony algorithm (IABC) based facial
expression recognition. In this proposed work, the face image is detected by Viola-Jones approach. By
using low pass filter the noises in the images are removed. Then the active facial patches with respect to
the position of eyes, eyebrows, nose, and lip corner are extracted from preprocessed images. In order to
obtain salient patches, the active patches are further processed. After that an optimal features are selected
by using IABC algorithm. Finally the different facial expression anger, disgust, fear, happiness, sadness,
and surprise are classified by using multiclass Support Vector Machine (MSVM). The experimental
results show that the proposed system achieves better performance compared with the existing system in
terms of accuracy, precision and recall.