Facial expression recognition by using intelligent systems and system automation

Facial expressions are described as the motion of eye, eyebrow, mouth and face to express a sentiment or an idea. Human beings do not have a problem to understand them however it is very complicated procedure to make computer to understand these motions. In this thesis six fundamental facial expressions; surprise, fear, disgust, joy, anger and sadness are classified with smart systems. The first step of facial recognition is to find facial region on the image. For this process skin color based face detection method is used. After facial region is detection, to find the position of organs on the face, vertical and horizontal projection histogram method is used. Then, the template which is constituted by sixty-six characteristic points that describe the shape of the organs on the detected facial region is placed on the facial region. In the occurrence of facial expression, the change of organs on the face is tracked with the aid of these points. For each facial expression, templates are constituted from these sixty-six points. Active Shape Models are used to determine which template is more suitable for the facial expression and to relocate the characteristic point to face edge. In order to do face detection process real time or/and to accelerate the process, data cluster should be minimized without loss or/and with minimum loss. To achieve this aim, Principal Component Analysis is used. After reducation of data that will be processed, to classify facial expression one of the intelligent systems Hidden Markov Model is used.