Super-resolution Reconstruction for Facial Images Based on Local Principal Component Analysis

DOI

10.2991/icacsei.2013.62How to use a DOI?

Keywords

Super-resolution reconstruction, Hallucinating faces, Local Principal Component Analysis.

Abstract

In order to improve the ghosting effect appearing the reconstruction results which are obtained through applying Principle Component Analysis (PCA) on the whole images, a novel algorithm which is reconstructed through applying PCA on local image patches is proposed. The new method firstly proposes the image patch model with overlapping areas. Then the input low-resolution patches are projected on the sample patches through PCA. And the weights can be obtained. Furthermore, the corresponding high-resolution patches are linearly combined through these weights to output the fusion patches. Now the best results are 16×12 reconstructions with the magnification of 8×8. Experimentations show that our method can reconstruct ultra-low resolution faces of 8×6 pixels with the magnification of 16×16, and the similarity with the original high-resolution images is higher.

Copyright

© 2013, the Authors. Published by Atlantis Press.

Open Access

This is an open access article distributed under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/).

Download article (PDF)

TY  - CONF
AU  - Jin ping He
AU  - Guang da Su
AU  - Jian sheng Chen
PY  - 2013/08
DA  - 2013/08
TI  - Super-resolution Reconstruction for Facial Images Based on Local Principal Component Analysis
BT  - Proceedings of the 2013 International Conference on Advanced Computer Science and Electronics Information (ICACSEI 2013)
PB  - Atlantis Press
SP  - 249
EP  - 252
SN  - 1951-6851
UR  - https://doi.org/10.2991/icacsei.2013.62
DO  - 10.2991/icacsei.2013.62
ID  - He2013/08
ER  -