and Technology University,China, " Face Image Super ResolutionviaAdaptive-Block PCA", The International Arab Journal of Information Technology (IAJIT) ,Volume 13, Number 06, pp. 16 - 23, November 2016, doi: .

Abstract: A novel single face imageSuper Resolution (SR)framework based onadaptive-blockPrincipal Component Analysis (PCA)is presented in this paper. The basic idea is the reconstruction of aHigh Resolution(HR)face image from aLow Resolution (LR)observation based on a set ofHR and LRtraining image pairs.TheHRimage blockis generated in the proposed method byusing the same position image blocks of each training image.The test face image and the training image setsare dividedinto many overlapping blocks,thenthese image blocks are classifiedaccording to the characteristics ofthe image blockand then PCAis operateddirectly on thenon-flatimage blocks to extract the optimal weightsandthe hallucinated patches are reconstructed using the same weights.The finalHRfacial image is formed by integrating the hallucinated patches. Experiments indicate that the new method producesHRfaces of higher quality and costs less computational time than some recent face imageSRtechniques.
URL: https://iajit.org/paper/4700

@ARTICLE{4700,
author={and Technology University,China},
journal={The International Arab Journal of Information Technology (IAJIT)},
title={Face Image Super ResolutionviaAdaptive-Block PCA},
volume={13},
number={06},
pages={16 - 23},
doi={},
year={1970}

,abstract={A novel single face imageSuper Resolution (SR)framework based onadaptive-blockPrincipal Component Analysis (PCA)is presented in this paper. The basic idea is the reconstruction of aHigh Resolution(HR)face image from aLow Resolution (LR)observation based on a set ofHR and LRtraining image pairs.TheHRimage blockis generated in the proposed method byusing the same position image blocks of each training image.The test face image and the training image setsare dividedinto many overlapping blocks,thenthese image blocks are classifiedaccording to the characteristics ofthe image blockand then PCAis operateddirectly on thenon-flatimage blocks to extract the optimal weightsandthe hallucinated patches are reconstructed using the same weights.The finalHRfacial image is formed by integrating the hallucinated patches. Experiments indicate that the new method producesHRfaces of higher quality and costs less computational time than some recent face imageSRtechniques.},
keywords={SR,face image, adaptive-block, PCA},
ISSN={2413-9351},
month={Jan}}

TY - JOUR
TI - Face Image Super ResolutionviaAdaptive-Block PCA
T2 -
SP - 16
EP - 23
AU - and Technology University
AU - China
DO -
JO - The International Arab Journal of Information Technology (IAJIT)
IS - 9
SN - 2413-9351
VO - 13
VL - 13
JA -
Y1 - Jan 1970
ER -
PY - 1970

AB - A novel single face imageSuper Resolution (SR)framework based onadaptive-blockPrincipal Component Analysis (PCA)is presented in this paper. The basic idea is the reconstruction of aHigh Resolution(HR)face image from aLow Resolution (LR)observation based on a set ofHR and LRtraining image pairs.TheHRimage blockis generated in the proposed method byusing the same position image blocks of each training image.The test face image and the training image setsare dividedinto many overlapping blocks,thenthese image blocks are classifiedaccording to the characteristics ofthe image blockand then PCAis operateddirectly on thenon-flatimage blocks to extract the optimal weightsandthe hallucinated patches are reconstructed using the same weights.The finalHRfacial image is formed by integrating the hallucinated patches. Experiments indicate that the new method producesHRfaces of higher quality and costs less computational time than some recent face imageSRtechniques.