Positional Ternary Pattern (PTP): An Edge Based Image Descriptor for Human Age Recognition
This paper presents a new edge based local feature descriptor, for age group recognition, named Positional Ternary Pattern (PTP). PTP code inherits the craniofacial shape, along with wrinkle and micro texture pattern, which proves its preeminence in aging process.
PROJECT OUTPUT VIDEO:
For representation of face aging, the active methods are: Anthropometric Model, Active Appearance Model (AAM), Aging Pattern Subspace (AGES), Aging Manifold, and Appearance Based Method.
Typical anthropometric models use geometric relationships between the facial components, like, eye, nose, mouth, jaw etc to describe face image. This approach can describe the craniofacial growth of face efficiently by using a few features, but fails to represent skin aging including wrinkle, spot and skin condition. AAM represents the shape and texture information at a time, which solves the misalignment problem of geometric feature based approach. However, AAM features do not capture wrinkle and skin information due to the dimensionality reduction made by Principle Component Analysis (PCA).
AGES is introduced to model the aging pattern, which is defined as the sequence of a particular individual’s face images sorted in time order, by constructing a representative subspace. Since, AGES assumes that, there are available images of the same individual at different ages, this method suffers from lack of dataset satisfying this criteria.
DISADVANTAGES OF EXISTING SYSTEM:
Aging manifold representation suffers from the lack of large number of training instances, that is required to learn the embedded manifold with statistical sufficiency.
In this paper, we propose a novel appearance based method, Positional Ternary Pattern (PTP), as edge based one for automatic age recognition with the notion that, both shape (the craniofacial growth) and skin (wrinkles and blemishes) changes during aging are efficiently detected on regions of face image with high edge response.
Our method is not only robust to noise but also describes face aging accurately and efficiently owing to adapting edge operators for generating patterns. In addition, our proposal achieves more stability than existing edge based methods by incorporating a ternary pattern which provides high discriminating power in flat and high-textured area.
We visualize the performance of the PTP with Support Vector Machine (SVM) on a challenging dataset and result shows that, PTP achieves better performance over existing methods.
ADVANTAGES OF PROPOSED SYSTEM:
We introduce a ternary pattern of the primary direction, which distinguishes the flat and edge-based region.
System : Pentium Dual Core.
Hard Disk : 120 GB.
Monitor : 15’’ LED
Input Devices : Keyboard, Mouse
Ram : 1GB.
Operating system : Windows 7.
Coding Language : MATLAB
Tool : MATLAB R2013A
Md Tauhid Bin Iqbal, Byungyong Ryu, Gihun Song and Oksam Chae, “Positional Ternary Pattern (PTP): An Edge Based Image Descriptor for Human Age Recognition”, IEEE International Conference on Consumer Electronics, 2016