Pain Facial Expression Recognition among 8 Facial Expressions using Similarity Measures

Facial expression recognition is one of the hottesttopics in machine vision domain. Lots of techniques and methodshave been tested for the recognition of six basic facial expressionsincluding happiness, surprise, disgust, anger, fear and sadness. Inthis paper, we focus on pain facial expression recognition,specifically. We propose a method in which facial features arefirst extracted and specific points are placed on the faceaccording to the result of facial feature extraction process.Displacements of specific points across different frames carry theinformation about which facial expression is taking place. Byclassifying the movements of specific points by an improvedversion of Euclidean distance similarity measure, facialexpression can be classified and recognized. Results prove theaccuracy of 90% on STOIC database.

کلیدواژه ها:

pain ، facial expression ، similarity measure tracking ، specific points