Face Editing Method Based On Pixel-level Face Labeling

With the rapid development of computer vision and computer graphics,the image has become the main carrier of people's daily information exchange because of its intu-itive and convenient transmission.Among them,the character image has been widely used in many fields,such as authentication,human-computer interaction,monitoring,forensics,medical diagnosis and treatment,prompting it to become one of the main top-ics in machine recognition and computer vision.Moreover,the face is the part of the human body contains the largest differential information,so people can get rich infor-mation only by observing the face,such as human gender,race,age,health,mood,and even occupation and so on.This paper is focus on face editing.With the popularity of mobile devices such as mobile phones,WeChat and mi-croblogging applications to promote the further development of social networks.Many people like to send photos of themselves or their families to social networks to get more attention.However,for the general user,their photographs are often lack of layering,or lack of exposure,or low color saturation.If there is no certain camera skills,it is difficult to shoot out the artistic conception and the beautiful photos.For some people who have "flaws" in the face,such as injuries,freckles,round face and so on.They are often difficult to cover these problems through make-up,and can only through the image editing methods to get a more satisfactory picture.Many of the existing image editing software,such as Adobe's Photoshop,Meitu Xiu Xiu and so on.Some of them can really achieve the effect of real ones,but it takes a certain amount of time for professional training,the general user is difficult to quickly get started.Although there are some software on the market can be a one-click fool operation,but the adjustment of the image does not retain the details,people can see a trace of the dressing.The existence of these problems makes the user's needs not fully satisfied,we need to further optimize.In order to solve the above problems and obtain the more realistic face image edit-ing effect,this paper obtains the high precision result of the face pixel information clas-sification based on the deep learning,and on the basis of the above results,we can get a more authentic and credible face editing image.This paper is mainly divided into two aspects:First,training fully convolution neural networks model,enter a face image,you can get the corresponding high precision facial pixel information classification re-sults.Second,automatic application in face editing operations,such as face smoothing,face cloning and face blending.With the help of the classification results,each piece of semantic area of the face can be treated separately,thus preserving the more attention to the details of the visual.The core of this paper is based on the fully convolution neural network.For the special topological structure of the face,the shape priors are added to obtain the approximate position of each type of label.Furthermore,it provides guidance for the learning of neural network.A large number of experimental results show that proposed algorithm not only supports the face image of any size and obtains high-precision pixel-level face image labeling results at a faster speed,but also obtains a convincing visual effect in face image editing.