Eigenlights: recovering illumination from face images

Authors:
James Burnstone;Hujun Yin
Affiliations:
School of Electrical and Electronic Engineering, The University of Manchester, Manchester, UK;School of Electrical and Electronic Engineering, The University of Manchester, Manchester, UK
Venue:
IDEAL'11 Proceedings of the 12th international conference on Intelligent data engineering and automated learning
Year:
2011

Citing 9
Cited 0

What Is the Set of Images of an Object Under All Possible Illumination Conditions?

International Journal of Computer Vision
Analytic PCA Construction for Theoretical Analysis of Lighting Variability in Images of a Lambertian Object

IEEE Transactions on Pattern Analysis and Machine Intelligence
Lambertian Reflectance and Linear Subspaces

IEEE Transactions on Pattern Analysis and Machine Intelligence
Face Identification across Different Poses and Illuminations with a 3D Morphable Model

FGR '02 Proceedings of the Fifth IEEE International Conference on Automatic Face and Gesture Recognition
Acquiring Linear Subspaces for Face Recognition under Variable Lighting

IEEE Transactions on Pattern Analysis and Machine Intelligence
Face Recognition from a Single Training Image under Arbitrary Unknown Lighting Using Spherical Harmonics

IEEE Transactions on Pattern Analysis and Machine Intelligence
Eigenfaces for recognition

Journal of Cognitive Neuroscience
Face Relighting from a Single Image under Arbitrary Unknown Lighting Conditions

IEEE Transactions on Pattern Analysis and Machine Intelligence
A Subspace Model-Based Approach to Face Relighting Under Unknown Lighting and Poses

IEEE Transactions on Image Processing

Quantified Score

Hi-index	0.00

Visualization

Abstract

In this paper, we present the use of subspace modelling to find the basis features of illumination across human face images. Instead of using a real image data set we use computer-generated 3D models, which we have built. Using these models we can better investigate the effect of the recognition of faces under illumination not confined within a particular trained subset. With this we have designed a recognition system where we investigate how many training images are necessary to build an illumination subspace that gives robust recognition. We aim to apply this technique to deal with the lighting problem in face recognition on mobile devices where some current methods are simply too complex to use.