Adaptive Detection and Tracking using Multimodal Information - Ciaran O Conaire (PhD Thesis)

Sep 11, 2026

Uploaded by Lute

This thesis investigates adaptive detection and adaptive object tracking within automated vision scenarios using fused multimodal information sources. A new adaptive object detection paradigm is developed via dynamic parameter selection, choosing detection thresholds to maximise agreement between independent pairs of sources. For tracking, a multi-source robust framework named the Spatiogram Bank tracker is proposed to address limitations of traditional histogram tracking. The work combines theoretical analysis with example applications using thermal infrared, visible spectrum video, and other multimodal inputs.

LanguageEnglish

CategoryPresentations

FormatPDF

Pages268

Reading Time1 h 34 min

AccessFree to download

1 Introduction

1.1 Thesis overview

2 Related Research

2.1 Introduction

2.2 Thermal Infrared

2.2.1 Thermal imaging technology

Adaptive Detection and Tracking using Multimodal Information Ciar¶an ¶O Conaire, B.Eng. School of Electronic Engineering & The Centre for Digital Video Processing Dublin City University Supervisor: Dr. Noel E. O'Connor A thesis submitted for the degree of Doctor of Philosophy September 2007 I hereby certify that this material, which I now submit for assessment on the programme of study leading to the award of Doctor of Philosophy is entirely my own work and has not been taken from the work of others save and to the extent that such work has been cited and acknowledged within the text of my work. Signed: ID No.: 99426439 Date: Ba mhaith liom an tr¶achtas seo a thiomni¶u do mo thuist¶³ Rh¶oda agus Breand¶an ¶O Conaire Bh¶³os an-¶og nuair a rugadh m¶e, ach bh¶³ an t-¶adh orm, mar thugadar aire agus gr¶a dom, agus n¶³ d¶oigh liom go mbeidh a leith¶eid¶³ ar¶³s ann. Acknowledgements There are a number of people to whom I am greatly indebted for their help, inspiration and patience. First of all I would like to acknowledge and thank my supervisor, Dr. Noel O'Connor, for his guidance, his advice and his ability to quickly resolve any issues I sought his help on. I would also like to thank him for giving me the freedom and the time to pursue the many paths this research has followed. I would like to thank the Irish Research Council for Science, Engineering and Technology (IRCSET) and Science Foundation Ireland (SFI) for providing the funding to pursue this PhD work. I would like to express my gratitude to my colleagues in the Centre for Digital Video Processing, and especially to Tomasz Adamek, Sorin Sav, Philip Kelly, Mike McHugh, Noel Murphy and Alan Smeaton. It has been a pleasure to work within the group with these people on so many interesting projects. And ¯nally, to all those people who helped keep me sane for the past year, a big thank you to Michael Cotter, Charlie Daly, Marian Flanagan, Susi Ghassempur, Ailbhe Jordan, Colm Mac Oireachtaigh, John Monaghan, Declan Nolan, Aisling N¶³ Chonaire, Cormac ¶O Conaire, Aonghus ¶O Conaire, Lisa O'Reilly, Karl Podesta and Sudanthi Wijewickrema. Abstract This thesis describes work on fusing data from multiple sources of information, and focusses on two main areas: adaptive detection and adaptive object tracking in automated vision scenarios. The work on adaptive object detection explores a new paradigm in dynamic parameter selection, by selecting thresholds for object detection to maximise agreement between pairs of sources. Object tracking, a complementary technique to object detection, is also explored in a multi-source context and an e±cient framework for robust tracking, termed the Spatiogram Bank tracker, is proposed as a means to overcome the di±culties of traditional histogram tracking. As well as performing theoretical analysis of the proposed methods, speci¯c example applications are given for both the detection and the tracking aspects, using thermal infrared and visible spectrum video data, as well as other multi-modal information sources. List of publications ² C. ¶O Conaire, N. E. O'Connor, and A. F. Smeaton. Detector adaptation by maximising agreement between independent data sources. In IEEE International Workshop on Object Tracking and Classi¯cation Beyond the Visible Spectrum, June 2007 . ² C. ¶O Conaire, N. E. O'Connor, and A. Smeaton. Thermo-visual feature fusion for object tracking using multiple spatiogram trackers. Journal of Machine Vision and Applications, May 2007 . ² C. ¶O Conaire, N. E. O'Connor, and A. F. Smeaton. An improved spatiogram similarity measure for robust object localisation. In IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP), April 2007 . ² C. ¶O Conaire, N. E. O'Connor, A. Smeaton, and G. J. F. Jones. Organising a daily visual diary using multi-feature clustering. In Proc. of 19th annual Symposium on Electronic Imaging, Jan 2007 . ² C. ¶O Conaire, N. E. O'Connor, E. Co

FAQs

What problem does the thesis address?

The thesis focuses on fusing data from multiple information sources to support adaptive detection and adaptive object tracking in automated vision scenarios.

How is adaptive object detection performed?

It uses a dynamic parameter selection approach that selects detection thresholds to maximise agreement between pairs of independent data sources.

What tracking method is proposed?

It proposes the Spatiogram Bank tracker, a multi-source robust tracking framework intended to overcome challenges of traditional histogram tracking.

Sep 11, 2026

Uploaded by Lute