a survey of face recognition technology wei-yang lin may 07, 2003
TRANSCRIPT
A survey of Face Recognition A survey of Face Recognition TechnologyTechnology
A survey of Face Recognition A survey of Face Recognition TechnologyTechnology
Wei-Yang LinWei-Yang LinMay 07, 2003May 07, 2003
Road Map• Introduction• Challenge in Face Recognition
– variation in pose– Variation in illumination
• Some recently works in FRT• Discussion
Introduction• FRT is a research area spanning
several disciplines.• Depending on the specific
application, FRT has different level of difficulty.
Challenges in FRT• The recent FERET test has
revealed that there are at least two major challenges:– The illumination variation problem– The pose variation problem
Illumination variation• Images of the same face appear
differently due to the change in lighting
• Naive Solution:– discarding the first few eigenfaces
Pose Variation• Basically, the existing solution can
be divided into three types:– multiple images in both training
stage and recognition stage– multiple images in training stage, but
only one image in recognition stage – single image based methods
Shape-from-Shading • The basic idea of SFS is to infer
the 3D surface of object from the shading information in image.
• Lambertian model has been used extensively in computer vision community for the SFS problem.
SFS results
Illumination cone• Illumination cone is a subspace
covers the variation in illumination.
Basis images
Synthetic images
Linear Object Class• How can we
recognize a face under different pose or expression when only one picture is given?
Linear Object Class
Curvature-based FRT• Use the curvature of surface to
perform face recognition • This is a great idea since the value
of curvature at a point on the surface is invariant under the variation of viewpoint and illumination
Elastic Bunch Graph• use Gabor wavelet
transform to extract face features so that the recognition performance can be invariant to the variation in poses.
2D-3D Face Recognition
• Almost all existing systems rely on either 2D images or 3D range data.
• 3D shape can compensate for the lack of depth information in 2D image.
• Therefore, integrating 2D and 3D information will be a possible way to improve the recognition performance.
ComparisonsInput Database
(images / person)
Applicable to existing image database
SFS One image One image All
Illumination cone
One image Several images Some
Linear object One image Several images Some
Curvature-based Range data Range data No
Elastic Bunch Graph
One image Several images Some
2D-3D combined Range data + one image
Range data + several images
No