Age estimation using local matched filter binary pattern
Imad Mohamed Ouloul, Karim Afdel, Abdellah Amghar · 2016
The automatic age estimation systems from facial images are often very complex and difficult to achieve, because of the aging process complexity. Such a system can be used in security, man-machine interactions, and biometrics. Research in this field has advanced during the last years. This paper proposes an age estimation system based on the shape and gray level texture intensity, extracted from facial images. The main contribution of this work is the design of a new descriptor named Local Matched Filter Binary Pattern, which detects and encodes face areas containing wrinkles. This descriptor, combined with parameters extracted by the active appearance model, enables the design of a high discriminative age. The experiments performed on the FG-net database, and the compared results approved this descriptor's efficiency in age estimation.