Face Skin Disease Detection with Textural Feature Extraction

Abstract

This study aims to build a model for the detection of facial skin diseases by utilizing the texture features in digital images of facial skin. The model is an automatic initial screening system for facial skin that can be used before carrying out further diagnosis processes by utilizing relatively expensive medical technology. Characteristics in facial images are obtained by extracting the textural features of the face digital image. Texture characteristics will distinguish the class of each facial problem based on their respective severity. The method used to extract textural features is the Gray Level Co-Occurrence Matrices (GLCM) method with the K-Nearest Neighbor classification method. The facial image data used were 150 digital images of problematic faces which were divided into 70% training data and 30% test data. This study produces a model accuracy of 80% accuracy with an error rate of 20%.