Abstract:
Automatic age prediction system for grayscale facial images is proposed in this paper. Ten
age groups, including, are used in the prediction
system. The process of the system is divided into
three phases: location, feature extraction, and
age prediction. Principal Component Analysis
(PCA) was used to reduce dimension and
enhance class. Finally Euclidean distance was
used to classify the images into one of seven
major groups. These groups are: Group1 (0 to
10 years), Group2 (11 to 20 years), Group3 (21
to 30 years), Group4 (31 to 40 years), Group5
(41 to 50 years), Group6 (51 to 60 years) and
Group7 (60 over). The proposed system is
experimented with 1300 facial images on a Core
2 Duo processor with 2 GB RAM. One half of the
images are used for training and the other half
for test. It takes 0.2 second to classify an image
on an average. The identification rate achieves
95.5% for the training images and 85.5% for the
test images, which is roughly close to human’s
subjective prediction.
