Please use this identifier to cite or link to this item:
Title: Prediction of Individual De-Aged Faces
Authors: Perera, C.N.I.
Keywords: Face de-aging
Research Subject Categories::TECHNOLOGY
Issue Date: 12-Sep-2013
Abstract: Human facial analysis has received significant attention and it has led to development of new means of performing errands such as face recognition, facial expression characterization, face modeling, and etc. In identifying aging patterns, although there are many research have been done in predicting forward aging pattern, we rarely can find researches which are done in the reversed direction considering both texture and color of skin. That is given an elderly picture of a person’s face, predicting the younger face. Therefore in this project, I’m trying to address the above reverse scenario also taking both color and texture into consideration. Age changes cause major variations in the appearance of human faces. Due to many lifestyle factors, it is difficult to precisely predict how individuals may look with advancing years or how they looked with "retreating" years. In this thesis, a new automatic aging scheme is proposed. This will not consider facial appearance changes due to health problems, operations, accidents and hair. It will consider normal biological aging process of a human. Simple tool have been developed to demonstrate the above de-aged face prediction. Tool takes input image with required age, current age, aging parameters and face feature points as input, and outputs the de-aged image. To do this transformation mathematical techniques with image processing techniques are used. Several faces have been used to depict the results. Results shows that faces when the age difference is high some face features are not very well de-aged. But the importance of this tool is it is very easy, simple to de-age old faces by not only changing the facial features but also adjusting the texture information.
Appears in Collections:Master of Computer Science - 2013

Files in This Item:
File Description SizeFormat 
  Restricted Access
2.85 MBAdobe PDFView/Open Request a copy

Items in UCSC Digital Library are protected by copyright, with all rights reserved, unless otherwise indicated.