Elkarazle, K.; Raman, V.; Then, P. Facial Age Estimation Using
Download 0.59 Mb. Pdf ko'rish
|
BDCC-06-00128
- Bu sahifa navigatsiya:
- Active appearance models (AAM)
- Ageing pattern subspace (AGES)
Texture-based models
: Unlike Anthropometric models, which use the distance be- tween facial points, texture-based models depend on the texture of facial images, which the pixel intensity can represent. In texture-based models, various image texture operators such as local binary patterns [ 29 ] (LBP) or biologically inspired features [ 30 ] (BIF) are employed to extract skin areas such as spots, lines, or edges that might represent wrinkles. Several studies, such as [ 31 , 32 ], used texture-based models to estimate age. Unlike anthropometric models, texture-based models can perform well on images taken in uncontrolled conditions; however, they are incapable of discriminating different shapes and distances between facial points. 3. Active appearance models (AAM) : Active appearance models are statistical mod- els commonly used in facial image representations, combining anthropometric and texture-based model descriptors. A dimensionality reduction algorithm such as the principal component analysis (PCA) learns the extracted texture and shape features. AAMs are ubiquitous in various facial recognition, facial verification, and age esti- mation tasks due to their flexibility in working with textures and shapes. However, reducing the feature’s dimensionality results in having several ageing features, such as wrinkles going unnoticed. Studies such as [ 33 , 34 ] used AAMs with various other methods to predict age. 4. Ageing pattern subspace (AGES) : In a study conducted by [ 35 ], ageing pattern sub- space (AGES) was proposed to identify a person’s ageing pattern based on a set of facial images taken at the age of 2, 4, and 8 sorted in ascending order. The reason for placing images in this order is to learn the ageing pattern of an individual, defined as a “sequence of personal face images sorted in time order” according to [ 36 ]. This method generates missing age samples by learning the subspace representation of a single image when constructing a sequence of the subject’s ageing facial images. AGES can help estimate missing samples; however, it does not work well with wrin- kles; therefore, integrating texture-based models with AGES is common. Studies such as [ 37 , 38 ] have used AGES by itself to extract facial features from photos. While studies such as [ 33 , 34 ] used AAMs with various other methods to predict age. Download 0.59 Mb. Do'stlaringiz bilan baham: |
Ma'lumotlar bazasi mualliflik huquqi bilan himoyalangan ©fayllar.org 2024
ma'muriyatiga murojaat qiling
ma'muriyatiga murojaat qiling