Facial Skin Defect Detection And Evaluation System

With the rapid development of medical beauty technology,people are paying more and more attention to the health of their skin.How to objectively and quantitatively evaluate the skin quality is also a key issue in the current medical beauty industry.The traditional large-scale mechanical skin detector has been unable to meet people's needs in terms of portability and real-time performance.The use of image processing methods for detecting and evaluating human skin has a very broad application prospect.This project hopes to establish a skin defect detection and evaluation system that can evaluate skin problems such as wrinkles,spots,acne,and pores on the skin surface.The specific work content is as follows:1.The background and research status of skin detection technology are introduced in detail.The common detection indexes for skin condition are analyzed.The skin defect is selected as the research object,and the face skin image database is constructed.In this paper,facial skin images were collected of 5 parts in 50 volunteers,10 pieces were collected from each part,and establish a database of 2500 original skin images.2.In view of the disadvantages of single feature on facial skin expression,this thesis proposes a feature fusion algorithm for skin detection.Firstly,studied in the texture features,shape features and color features,and the LBP features,wavelet features,HOG features and color moment features are selected to describe facial skin defects.By analyzing the above-mentioned features on the detection of facial skin defects,the rotation-invariant equivalent LBP features,HOG features,and color moment features are merged.Finally,we employ SVM method to train detection model.Extensively experimental results show that the proposed method achieves better detection performance than single feature.Considering the shortage of manual designed facial skin feature,this thesis proposes a facial skin feature detection approach based on convolutional neural network.Designed and implemented the VGG16-based method for detecting facial skin defects.Specifically,the VGG16 network model is first constructed,and then the self-built skin image database is used for training to fine-tune the parameters in the network.The "batchsize" parameter of the VGG16 classifier is set to 20,and the "max_batches" parameter is 64.The recognition rate is as high as 97.66%,which improves the overall recognition performance of the system to a large extent.4.Design a facial skin defect evaluation model.This paper combines the classification criteria of skin acne,pores and wrinkles in the domestic medical community,and classifies all kinds of skin imperfections based on the size of the defect area.According to the grades of various types of cockroaches,the overall evaluation model of face skin was proposed,and the human-computer interaction interface was designed to complete the construction of the whole face skin flaw detection and evaluation system.