Interpolation between eigenspaces using rotation in multiple dimensions

Authors:
Tomokazu Takahashi;Lina Lina;Ichiro Ide;Yoshito Mekada;Hiroshi Murase
Affiliations:
Graduate School of Information Science, Nagoya University, Japan and Japan Society for the Promotion of Science;Graduate School of Information Science, Nagoya University, Japan;Graduate School of Information Science, Nagoya University, Japan;Department of Life System Science and Technology, Chukyo University, Japan;Graduate School of Information Science, Nagoya University, Japan
Venue:
ACCV'07 Proceedings of the 8th Asian conference on Computer vision - Volume Part II
Year:
2007

Citing 3
Cited 0

Illumination Planning for Object Recognition Using Parametric Eigenspaces

IEEE Transactions on Pattern Analysis and Machine Intelligence
From Few to Many: Illumination Cone Models for Face Recognition under Variable Lighting and Pose

IEEE Transactions on Pattern Analysis and Machine Intelligence
Principal Manifolds and Bayesian Subspaces for Visual Recognition

ICCV '99 Proceedings of the International Conference on Computer Vision-Volume 2 - Volume 2

Quantified Score

Hi-index	0.00

Visualization

Abstract

We propose a method for interpolation between eigenspaces. Techniques that represent observed patterns as multivariate normal distribution have actively been developed to make it robust over observation noises. In the recognition of images that vary based on continuous parameters such as camera angles, one cause that degrades performance is training images that are observed discretely while the parameters are varied continuously. The proposed method interpolates between eigenspaces by analogy from rotation of a hyper-ellipsoid in high dimensional space. Experiments using face images captured in various illumination conditions demonstrate the validity and effectiveness of the proposed interpolation method.