We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
Age progression by gender-specific 3D aging model.
- Authors
Riaz, Sidra; Park, Unsang; Choi, Jongmoo; Natarajan, Prem
- Abstract
Facial aging is an important problem of face recognition in missing children and automatic template update. As aging is a temporal process, it alters the facial appearance of the individuals. The sources of variations in facial appearance are caused by wrinkles (under eyes, forehead, around lips, and jawline), facial growth (cranial size and skull), and skin tone. The other factors such as health, lifestyle, and gender also impose variations in the aging process. Therefore, predicting facial aging with considering all those factors is a very difficult task. We present our 3D gender-specific aging model which automatically produces simulated images at age y by taking only one input image at age x irrespective of the pose and lighting conditions. The gender-specific aging model is constructed by various datasets (FG-NET, PCSO, Celebrities, BROWNS, Private), and its quality is evaluated with respect to various combinations of the datasets. We further fine-tune the aging model by changing the length of shape and texture eigenvectors and examine how these parameters affect the simulation results. Comparisons of the simulation results with state-of-the-art approaches as well as ground truth images demonstrate the effectiveness of the proposed methods. The subjective and objective evaluations are also carried out which emphasize the potential of our proposed gender-specific 3D aging model.
- Subjects
HUMAN facial recognition software; WRINKLES (Skin); EIGENVECTORS; SIMULATION methods &; models; RENDERING algorithms
- Publication
Machine Vision & Applications, 2019, Vol 30, Issue 1, p91
- ISSN
0932-8092
- Publication type
Article
- DOI
10.1007/s00138-018-0975-2