Pain recognition and intensity classification using facial expressions

Facial biometrics, specifically facial expression analysis, is one of the most actively investigated topics towards the creation of an automated system capable of detecting and classifying pain in human subjects. This paper presents a comparative analysis of Gabor energy filter based approaches combined with powerful classifiers, such as Support Vector Machines, for pain detection and classification into three levels. The intensity of pain is labelled using the Prkachin and Solomon Pain Intensity scale. In this paper, the levels of intensity have been quantized into three disjoint groups: no pain, weak pain and strong pain. The results of experiments show that Gabor energy filters provide comparable or better results compared to previous filter-based pain recognition methods, with a 74% classification rate of pain versus no pain, and 74%, 30% and 78% precision rates when distinguishing pain into no pain, weak pain and strong pain respectively.