Image based analysis of emotional expression development in facial transplant patient

In this study, in healthy groups and facial transplantation patient, Gabor Wavelet transform based on appearance was adopted for recognition neutral and 6 emotional facial expressions consisting of angry, scared, happy, hate, confused and sad. The aim is to examine the ratio of regional performing facial mimics of face transplantation patient acording to healthy groups during each expressions. The system consists two phases; in the first stage, facial images of the healthy group were used for the training and test input, and in the second stage, the images of facial transplantation patient were used for the test input of trained system. Feature vectors were obtained from face images which convolved with Gabor filter (5-scales, 8-orientations) banks to feature extraction from the location of the eyes and mouth which is the most prominent region, while performing expressions. This feature vectors were classified by K-nearest neighbor classifier. The results were evaluated for normal groups and transplant patients.