Facial expression recognition by applying multi-step integral projection and SVMs

In order to achieve subject-independent facial feature detection and extraction and obtain robustness against illumination variety, a novel method of facial expression recognition using the combination of multi-step integral projection and Gabor transformation for feature detection and SVM for classification is presented in this paper. First, to avoid manually picked expression features, we propose a new approach called multi-step integral projection to detect and locate the exact position of human facial features automatically. Second, we segment the extracted areas into small cells for 7×7 pixels each and apply Gabor transformation on each cell. This greatly reduces the execution time of the Gabor transformation while retaining important information. Third, a Support Vector Machine is used for classifying facial emotions and we tested our system on the JAFFE database while achieving a high recognition rate of 94.8357% on trained data. Finally, we discuss the effect of different parameters selection in Gabor transformation and analyze the reason for some incorrect recognition.