Facial Expression Recognition based on Support Vector Machine using Gabor Wavelet Filter

Face is the most important part of human body. Facial expression is a way of nonverbal communication with one another. Human face expresses the internal emotional feelings and contains important information. It is our goal to extract considerable features used for real-time Facial Expression Recognition (FER) system. Facial expression can be recognized by both facial shape features and appearance features. In our proposed methodology, we first extract the shape features from positions on a face. Then multi-orientation Gabor wavelet coefficient feature are extracted from expression images. We have used Support Vector Machines (SVM) as classifier. As face has some fixed special points, linear classifier works excellent on facial point data. Thus SVM performs with satisfactory outcomes in our FER system. Our experimental result shows that using facial shape features and Gabor wavelet coefficient based on SVM is more accurate and faster most other previously proposed methodologies.