Paper Title
FCHE: Face Classification Using Histogram Equalization
Abstract
Face recognition subject to uncontrolled-blur, illumination, occlusion errors are challenging and which gives
negative impact on face recognition accuracy. Hence, error removal is necessary in face recognition methods before the
application of other techniques for face classification. To achieve this we propose a new method called Face Classification
using Histogram Equalization and and morphological operations. Using mirror image technique the virtual images for both
training and test samples are generated. The conventional and inverse representation methods on these images are applied
separately further, they are combined to achieve better accuracy using weighted fusion. The experiments are carried out on
publicly available FERET, ORL, Extended YALE and JAFEE normal and corrupted face data bases. The results show that,
FCHE is an improvement over conventional and inverse representation based linear regression classification (CIRLRC). It is
demonstrated that, FCHE performs better than the other state-of-the-art conventional representation based Face classification
Methods.
Keywords- Face Recognition; Morphological functions; Linear Regression; Classification; Conventional Representation;
Inverse Representation.