Font Size: a A A

Research And Implementation Of The Face Beautification Technology

Posted on:2011-12-25Degree:MasterType:Thesis
Country:ChinaCandidate:P ZhaoFull Text:PDF
GTID:2178330332961428Subject:Communication and Information System
Abstract/Summary:PDF Full Text Request
In real life, most people do not meet perfect face standard. There are some defects, such as the face is too wide, too long or too narrow, too short. "Everyone has his heart in beautiful things", how deform ordinary face to perfect face is issue which people more concerned about. So the research of facial beatification deformation technology is very important.There are many facial deform algorithms, every algorithm has its strongpoint and weakness. Cross-decomposition has no space transform, the deformation results are not satisfactory. Distorted grid algorithm has space transform, but the features marked is very cumbersome, and the grid does not allow collapse. Lines facial deformation method can be deformed on purpose, but the computational complexity is very high, and very likely to lead "ghost" phenomenon. Thin plate splines and moving least square algorithm of the facial deformation technology is based on control points, has strong control ability, can produce many special effects and animation. So thin plate splines and the moving least square deformation algorithm is selected for facial deformation.Face image beautification software can make face beautiful, but the software are unable to provide perfect facial contours, so user only depends on the feeling and experience to deform, they can not deform the ordinary facial contours to perfect facial contours accurately. In order to overcome the defect of the software, the paper presents a method for ordinary facial contour deformation. Firstly, face beautiful evaluation criteria is proposed and the quantitative beautiful features of the face is achieved. Through the comparative analysis between ordinary facial feature vector and the standard value of perfect human face, the main factor of the imperfect face is imperfect facial contour. Secondly, the perfect face contour is extracted by using the GVF snake model and is target features for deformation. Finally, using moving least square and thin plate splines facial deformation algorithm is to realize the deformation from the ordinary facial contour to perfect facial contour. The results show that the method can overcome the human face image beautification software defects. The facial contour after deformation is very smooth and realistic. The face beauty rank is improved obviously.
Keywords/Search Tags:Facial deformation, Moving least square, Thin plate splines, Beautiful standard, GVF snake model
PDF Full Text Request
Related items