Human Skin Wrinkle Detection Using The Convolutional Neural Network Method

Abstract

Wrinkles are a visual indicator of skin aging and are widely used in dermatological and cosmetic assessments. However, automatic wrinkle detection from facial images remains challenging due to illumination variation, image noise, and subtle skin texture characteristics. This study applies a Convolutional Neural Network (CNN) for human skin wrinkle detection using image preprocessing techniques, including intensity normalization, Contrast Limited Adaptive Histogram Equalization (CLAHE), denoising, and sharpening. Experiments were conducted on 600 facial skin images obtained from publicly available sources and manually categorized into wrinkled and non-wrinkled classes. To ensure result reliability, the dataset was divided into training, validation, and testing sets using a 70:20:10 ratio. The experimental results show that the proposed approach achieved an accuracy of 0.9136, demonstrating consistent performance across validation and test sets.