A Facial Recognition Method Based On Dmw Transformed Partitioned Images

Abstract

A new approach based on applying the Two Dimensional Discrete Multiwavelet Transform (2D DMWT) to the partitioned faces is proposed in this paper for face recognition. First, the input facial image is divided into six parts in the preprocessing step to reduce the effect of the unnecessary information (background) on the system performance. Then, the 2D DMWT is applied to each part for feature extraction and dimensionality reduction. Finally, the l2-Norm is applied for further data compression. These features are classified using a Neural Network (NN) classifier. The proposed techniques are evaluated using four databases, namely, ORL, YALE, FERET, and FEI. These databases cover a wide variety of facial variations, such as facial expressions, rotations, illuminations, etc. K-fold Cross Validation is used to analyze the experimental results. The proposed system is shown to improve the recognition rates as well as reduce the storage requirements compared to the existing approaches.

Publication Date

9-27-2017

Publication Title

Midwest Symposium on Circuits and Systems

Volume

2017-August

Number of Pages

1352-1355

Document Type

Article; Proceedings Paper

Personal Identifier

scopus

DOI Link

https://doi.org/10.1109/MWSCAS.2017.8053182

Socpus ID

85034070204 (Scopus)

Source API URL

https://api.elsevier.com/content/abstract/scopus_id/85034070204

This document is currently not available here.

Share

COinS