Real-time facial feature point detection and tracking using rule-based approach and integral projection

Abstract

The face is a basis of how one can describe emotions of another as it is a medium of communication. Through looking at the face of the person, one can detect what kind of feeling he is portraying. It may be a case for humans; however, computers cannot detect the emotion of humans simply by looking at their faces, computers rely on the positioning of the feature points of the eyes, and mouth in order to determine the emotion. In this connection, this study aims to create a system wherein it can detect the features and track points of the face in a real-time video. The authors attempt to combine different algorithms in order to detect and track the feature points in the face. Based on the results of the conducted tests, the setup has recorded an overall success rate of 44.56 %. This is due to the large skin range that was used in the program.