Research On Intellident Face Skin Health Detection System

Skin disease is a common and multiple disease in medicine.In view of the increasing demand for beauty and skin care,skin detection technology has attracted more and more attention.Traditional manual diagnosis has certain subjectivity and cannot meet the requirements of complex and diverse pigmented skin disease detection.At present,for the problem of facial skin detection,domestic and foreign research is mainly focused on the beauty field of large professional equipment.Therefore,it is of great significance to design a system that can intelligently and accurately detect skin conditions and is simple in structure and suitable for the public to reflect facial skin health.This paper uses the combination of machine learning algorithms and portable smart terminal equipment to design a complete smart face and skin health detection system.The whole system includes two parts: software detection and hardware construction.In order to obtain the different features of pigmented skin disease under different light waves,a control system based on embedded combined light sources(daylight,ultraviolet,wood light,parallel polarized and cross-polarized)was designed.The ordinary camera mode of Android smart phone is used to replace the huge camera device for image acquisition under different light sources.The combination of the two greatly reduces the complexity of subsequent algorithm processing.The software detection part is combined with deep learning and improved capsule network(CapsNet)to classify and recognize the collected images.A skin detection model based on deep learning algorithm was built,and the improved CapsNet model was used to classify and recognize the preprocessed skin images.After classification,seven pigmented skin diseases such as melanoma,melanin cell nevus,keratosis/bowen disease and vascular lesions can be detected.Finally,The system will send the detection results to the Android phone APP,the users can realize on the Android phone face pigmentation skin disease detection,check the history of detection records,modify personal information.By comparing with the traditional convolutional neural network model,the improved CAPSNET model can achieve better recognition effect for pigmented skin disease,and the accuracy rate is 10% higher than that of the convolutional neural network model.Finally,the experimental verification shows that the five light modes of the system are of great help to the extraction of different features of the skin(Skin texture,vascular lesions,spots,etc),and can accurately identify the type of pigmented skin disease,which can achieve universal application to the public.