Facial expression recognition for neonatal pain assessment

Abstract

Facial expressions are considered a critical factor in neonatal pain assessment. This paper attempts to apply modern facial expression recognition techniques to the task of distinguishing pain expression from non-pain expression. Firstly, 2D Gabor filter is applied to extract the expression features from facial images. Then we apply Adaboost as a feature selection tool to remove the redundant Gabor features. Finally, the Gabor features selected by Adaboost are fed into the support vector machines (SVMs) for final classification. 510 facial images are investigated by using SVMs. The best recognition rates of pain versus non-pain (85.29%), pain versus calm (94.24%), pain versus cry (78.24%) were obtained from an SVM with a polynomial kernel of degree 3. The results of this study indicate that the application of SVM technique in pain assessment is a promising area of investigation.