Facial emotion recognition method based on Pyramid Histogram of Oriented Gradient over three direction of head

Today Human Computer Interaction (HCI) is one of the most important topics in machine vision and image processing fields. Through features can get beneficial information about the variety of emotions and gestures which are produced by the movements of facial major parts. In this paper we presented the technique of Pyramid Histogram of Oriented Gradient for feature extraction and compare it with gabor filters. Six basic facial expressions plus the neutral pose are considered in the evaluations. The KNN and SVM techniques are used in the classification phase. Unlike most emotion detection approaches that focus on frontal face view this method concentrates on three views of the face and can easily be generalized to other poses and feelings. We have tested our algorithm on the Radboud faces database (RaFD) over three directions of head (frontal, 45 degree to the right and 45 degree to the left). Cohn-Kanade (CK+) and JAFFE are two other databases used in this work. The experiments using the proposed method demonstrate favorable results. In the best condition by using Pyramid Histogram Of Oriented Gradient plus KNN classification, the success rates were 100, 96.7, 98.1, 98.3 and 98.9 % for RaFD (frontal pose), RaFD (45 degree to the right), RaFD (45 degree to the left), JAFFE and CK+ databases respectively.