Identifying the Human Age Group Using Active Appearance Model and KNN Classifier

The classification of human age group using facial images have a vital role in image processing and computer
vision. The task of facial processing is widely used in security and multimedia application. The human beings
are capable to categorize a person’s age group from the facial images. It is proved that computer can classify
the human age group according to facial features using Active Appearance Model (AAM) and Sammon’s
mapping method. So the challenge is to develop an age group classification system by using the different
methodologies. The existing system is based on facial images uses in Cost-Sensitive Ordinal Hyper Planes
Ranking Algorithm (CSOHP) with an effective descriptor Scattering Transform.Age group classification
method uses that describe Ageing pattern Subspace (AGES), Warped Gaussian Process (WGP), AdaBoost
algorithm and also Local Binary Pattern algorithm. Accuracy of classification is based on the algorithm used
for Extracting the feature points and Mis-classification is possible if the image quality is poor. The framework
is composed of four main modules (i) Pre-Processing, (ii) Feature Extraction using Active Appearance model
(AAM), (iii) Reducing the dimensions using Sammon’s mapping method, (iv) Classification of facial images
using K-Nearest Neighbor (K-NN).The proposed approaches uses four pre-processing techniques such as
cropping face image, Gamma Correction, DoG filtering & Contrast Equalization, then Extract the features are
extracted from the facial images using an Active Appearance Model(AAM) and the dimensions are reduced
further using Sammon’s Mapping method. The facial images are classified into the three different age group
Adolescence, Adult, and Senior Adult using K-Nearest Neighbour (k-NN) classifier algorithm, using color Feret
database. A maximum classification rate of 90% is achieved in using KNN classifier algorithm.