Classification of Human Skin Color and its Application to Face Recognition
Marwa Jmal, Wided Souidene Mseddi, Rabah Attia, Anis Youssef · 2014
Abstract—This paper presents a novel human skin color classification into skin color tones: White and Black. This is performed by developing a skin color classifier based on pixel-based classification using RGB model. Our proposed method is classified under the category of an explicitly defined skin region model. The skin classifier divides our database formed by some images from the FERET set of faces into two sub-databases according to the skin color. The skin color classification method is then applied on a face recognition technique by reducing the number of trained images in the matching process. The performance of the proposed human skin color classifier is evaluated perceptually. Experimental results showed that our proposed skin color classifier is able to classify a face into its possible skin color tone and reaches 87% as hit rate. Keywords-Skin color classification; Face recognition; Pixel-based classification; RGB model.