Open AccessFacial Age Estimation Using Spatial Weber Local Descriptor
Author(s) -
Asuman Günay,
Vasif V. Nabiyev
Publication year - 2017
Publication title -
international journal of advances in telecommunications, electrotechnics, signals and systems
Language(s) - English
Resource type - Journals
ISSN - 1805-5443
DOI - 10.11601/ijates.v6i3.218
Subject(s) - histogram , artificial intelligence , pattern recognition (psychology) , block (permutation group theory) , subspace topology , computer science , texture (cosmology) , feature extraction , estimation , mathematics , spatial analysis , feature (linguistics) , scale (ratio) , computer vision , image (mathematics) , statistics , geography , cartography , linguistics , philosophy , geometry , management , economics
This paper introduces a novel age estimation method using a new texture descriptor Weber Local Descriptor (WLD). This texture descriptor is analyzed in depth for age estimation problem. In the study, the multi-scale versions of holistic and spatial WLD (SWLD) descriptors are used to extract the age related features from normalized facial images. After finding a lower dimensional feature subspace, age estimation is performed using multiple linear regression. In addition the age estimation accuracy of each of the distinct and intersection block used in spatial texture extraction are investigated. Experiments on FGNET, MORPH and PAL databases have shown that similar age estimation performances can be obtained by using more effective blocks in spatial histogram generation. This also provides us to reduce the number of features and computational cost.