University of Surrey

Test tubes in the lab Research in the ATI Dance Research

ON AGGREGATION OF LOCAL BINARY DESCRIPTORS

Husain, S and Bober, M (2016) ON AGGREGATION OF LOCAL BINARY DESCRIPTORS In: 3rd IEEE International Workshop on Mobile Multimedia Computing (MMC 2016), 2016-07-11 - 2016-07-15, Seattle, USA.

[img]
Preview
Text
camera-ready_icme2016.pdf - Accepted version Manuscript
Available under License : See the attached licence file.

Download (1MB) | Preview
[img]
Preview
PDF (licence)
SRI_deposit_agreement.pdf
Available under License : See the attached licence file.

Download (33kB) | Preview

Abstract

This paper addresses the problem of aggregating local binary descriptors for large scale image retrieval in mobile scenarios. Binary descriptors are becoming increasingly popular, especially in mobile applications, as they deliver high matching speed, have a small memory footprint and are fast to extract. However, little research has been done on how to efficiently aggregate binary descriptors. Direct application of methods developed for conventional descriptors, such as SIFT, results in unsatisfactory performance. In this paper we introduce and evaluate several algorithms to compress high-dimensional binary local descriptors, for efficient retrieval in large databases. In addition, we propose a robust global image representation; Binary Robust Visual Descriptor (B-RVD), with rank-based multi-assignment of local descriptors and direction-based aggregation, achieved by the use of L1-norm on residual vectors. The performance of the B-RVD is further improved by balancing the variances of residual vector directions in order to maximize the discriminatory power of the aggregated vectors. Standard datasets and measures have been used for evaluation showing significant improvement of around 4% mean Average Precision as compared to the state-of-the-art.

Item Type: Conference or Workshop Item (Conference Paper)
Subjects : Electronic Engineering
Divisions : Faculty of Engineering and Physical Sciences > Electronic Engineering
Authors :
AuthorsEmailORCID
Husain, SUNSPECIFIEDUNSPECIFIED
Bober, MUNSPECIFIEDUNSPECIFIED
Date : 2016
Copyright Disclaimer : © 2016 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.
Uncontrolled Keywords : visual search, binary descriptors, global descriptor, image retrieval
Depositing User : Symplectic Elements
Date Deposited : 22 Jul 2016 10:20
Last Modified : 22 Jul 2016 10:20
URI: http://epubs.surrey.ac.uk/id/eprint/811329

Actions (login required)

View Item View Item

Downloads

Downloads per month over past year


Information about this web site

© The University of Surrey, Guildford, Surrey, GU2 7XH, United Kingdom.
+44 (0)1483 300800