Learning linear discriminant projections for dimensionality reduction of image descriptors

Cai, H, Mikolajczyk, K and Matas, J (2008) Learning linear discriminant projections for dimensionality reduction of image descriptors In: Proceedings of the British Machine Vision Conference, 2008-09-01 - 2008-09-04, Leeds, UK.

Preview	Text DimensionalityReduction_final.pdf Available under License : See the attached licence file. Download (556kB) | Preview
Preview	Text (licence) SRI_deposit_agreement.pdf Available under License : See the attached licence file. Download (33kB) | Preview

Abstract

This paper proposes a general method for improving image descriptors using discriminant projections. Two methods based on Linear Discriminant Analysis have been recently introduced in [3, 11] to improve matching performance of local descriptors and to reduce their dimensionality. These methods require large training set with ground truth of accurate point-to-point correspondences which limits their applicability. We demonstrate the theoretical equivalence of these methods and provide a means to derive projection vectors on data without available ground truth. It makes it possible to apply this technique and improve performance of any combination of interest point detectors-descriptors. We conduct an extensive evaluation of the discriminative projection methods in various application scenarios. The results validate the proposed method in viewpoint invariant matching and category recognition.

Item Type:	Conference or Workshop Item (Conference Paper)
Divisions :	Faculty of Engineering and Physical Sciences > Electronic Engineering > Centre for Vision Speech and Signal Processing
Authors :	Name Email ORCID Cai, H Mikolajczyk, K Matas, J
Date :	2008
Identification Number :	10.5244/C.22.51
Additional Information :	Copyright 2008 The Authors
Depositing User :	Symplectic Elements
Date Deposited :	15 Oct 2014 14:41
Last Modified :	31 Oct 2017 16:58
URI:	http://epubs.surrey.ac.uk/id/eprint/806142

Actions (login required)

View Item

Downloads