Two dimensional incremental reduction PCA: A novel appearance based technique for image representation and recognition

R. M. Mutelo, S. S. Dlay, W. L. Woo

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

2 Citations (Scopus)

Abstract

In this paper, an efficient two dimensional incremental reduction principal component analysis approach (2DIRPCA) is developed for image representation and recognition. The 2DIRPCA technique computes the image covariance matrix of each image matrix as it arrives sequentially. Therefore, the contribution of each image to the projection matrices is added to the existing projection matrices. In this way, the 2DIRPCA method overcomes the limitations such as the computational cost and memory requirements making it suitable for real time applications. The feasibility of the proposed approach was tested on the ORL database consisting of 400 face images. The 2DIRPCA method shows superior performance in terms of computational time, storage and recognition accuracy (93.5%) with a 10 × 6 feature matrix compared to the 2DPCA (92.5%) with a 112 × 7 feature matrix.

Original languageEnglish
Title of host publication5th International Conference on Visual Information Engineering, VIE 2008
PublisherIEEE
Pages588-593
Number of pages6
Edition543 CP
ISBN (Electronic)9780863419140
DOIs
Publication statusPublished - 9 Jan 2009
Event5th International Conference on Visual Information Engineering, VIE 2008 - Xi'an, China
Duration: 29 Jul 20081 Aug 2008

Conference

Conference5th International Conference on Visual Information Engineering, VIE 2008
Country/TerritoryChina
CityXi'an
Period29/07/081/08/08

Keywords

  • Face recognition
  • Feature extraction
  • Two dimensional principal component analysis (2DPCA)
  • Two dimensional reduction principal component analysis (2D RPCA)

Fingerprint

Dive into the research topics of 'Two dimensional incremental reduction PCA: A novel appearance based technique for image representation and recognition'. Together they form a unique fingerprint.

Cite this