Paper Info
Title
A robust facial image super-resolution model via mirror-patch based neighbor representation
Abstract
Many patch-based facial image super-resolution (or hallucination) techniques have been proposed in literature but all of them fail in the presence of high-density impulse noise and occlusion. A novel mirror-patch based neighbor representation (MPNR) model is proposed here which uses mirror-patch based data fidelity along with the input-patch based fidelity in low-resolution (LR) space to address the above problem. The computation of mirror-patch based data fidelity helps in compensating the corrupted features of an input patch through its mirror-patch. The objective function of the proposed model is designed in such a way that it hallucinate the input LR faces and takes care of occlusion/heavy noise effect simultaneously in the reconstruction process. It is conspicuous from experimental results attained on FEI and CAS-PEAL-R1 databases that the proposed MPNR model has outperformed all the comparative methods.
Year
DOI
Venue
2019
10.1007/s11042-019-07791-y
Multimedia Tools and Applications
Keywords
Field
DocType
Face super-resolution, Face hallucination, Position-patch based reconstruction
Computer vision,Fidelity,Face hallucination,Pattern recognition,Computer science,Impulse noise,Artificial intelligence,Superresolution,Computation,Hallucinate
Journal
Volume
Issue
ISSN
78
18
1380-7501
Citations 
PageRank 
References 
3
0.37
0
Authors
2
Name
Order
Citations
PageRank
Shyam Singh Rajput1222.61
K. V. Arya228928.09