Automatic scoring of wrinkles on the forehead

Abstract

Scoring facial wrinkles is a crucial process to specify exact treatment for wrinkles or to understand the effectiveness of the applied treatment to wrinkles. In this study, only forehead wrinkles are examined instead of the whole facial wrinkles. Detection and quantification of the forehead wrinkles are tried to be performed automatically by computers. For this purpose, the features are extracted using local binary pattern and Gabor filter which is based on connected component analysis. Facial images are labeled by educated volunteers according to their observation of wrinkle. By applying these two methods to labeled images, LBP achieved 64.5% accuracy whereas Gabor filter based on connected component analysis achieved 79.7% accuracy.