Cambridge University Engineering Department

	Department of Engineering

University of Cambridge > Engineering Department

Paul McIlroy

PhD student supervised by Professor Roberto Cipolla and Dr Ed Rosten
in the Machine Intelligence Laboratory at CUED.
Also a Munro Student and Part IA (Maths & Electrical) supervisor at Queens' College.
Intern at Microsoft Research Cambridge supervised by Dr Andrew Fitzgibbon

First two years of PhD supervised by Professor Tom Drummond

[C.V. | Résumé] updated November, 2012


	Kinectrack: Agile 6-DoF Tracking Using a Projected Dot Pattern McIlroy, Izadi and Fitzgibbon This paper proposes a new six degree-of-freedom (6-DoF) tracker which allows real-time and low-cost pose estimation using only commodity hardware. The dot pattern emitter and IR camera of the Kinect are separated. Keeping the camera fixed and moving the IR emitter in the environment, the 6-DoF pose of the emitter is recovered by matching the observed dot pattern in the field-of-view of the camera to a pre-captured reference image. A novel matching technique obtains dot pattern correspondences efficiently in wide- and adaptive-baseline scenairos. The system recovers piecewise planar 3D scene structure, enabling live 3D reconstruction in untextured environments. [2012 ISMAR Paper] [2012 ISMAR Video] This work won best paper runner up at ISMAR 2012.

Kinectrack: Agile 6-DoF Tracking Using a Projected Dot Pattern
McIlroy, Izadi and Fitzgibbon

This paper proposes a new six degree-of-freedom (6-DoF) tracker which allows real-time and low-cost pose estimation using only commodity hardware. The dot pattern emitter and IR camera of the Kinect are separated. Keeping the camera fixed and moving the IR emitter in the environment, the 6-DoF pose of the emitter is recovered by matching the observed dot pattern in the field-of-view of the camera to a pre-captured reference image. A novel matching technique obtains dot pattern correspondences efficiently in wide- and adaptive-baseline scenairos. The system recovers piecewise planar 3D scene structure, enabling live 3D reconstruction in untextured environments.

[2012 ISMAR Paper] [2012 ISMAR Video] This work won best paper runner up at ISMAR 2012.


	High-level Scene Structure using Visibility and Occlusion McIlroy, Cipolla and Rosten This paper proposes a new method for extracting high-level scene information from the type of data available from simultaneous localisation and mapping systems. The scene is modelled using a collection of primitives (such as bounded planes) and explicit use is made of both visible and occluded points in order to refine the model. Since the formulation allows for different kinds of primitives and an arbitrary number of each, Bayesian model evidence is used to compare very different models on an even footing. The results show that explicit reasoning about occlusion improves model accuracy and yields models which are suitable for aiding data association. [2011 BMVC Abstract] [2011 BMVC Paper]

High-level Scene Structure using Visibility and Occlusion
McIlroy, Cipolla and Rosten

This paper proposes a new method for extracting high-level scene information from the type of data available from simultaneous localisation and mapping systems. The scene is modelled using a collection of primitives (such as bounded planes) and explicit use is made of both visible and occluded points in order to refine the model. Since the formulation allows for different kinds of primitives and an arbitrary number of each, Bayesian model evidence is used to compare very different models on an even footing. The results show that explicit reasoning about occlusion improves model accuracy and yields models which are suitable for aiding data association.

[2011 BMVC Abstract] [2011 BMVC Paper]


	Deterministic Sample Consensus with Multiple Match Hypotheses McIlroy, Rosten, Taylor and Drummond This paper proposes a deterministic scheme for selecting correspondences from feature matching to generate motion hypotheses. The method combines matching scores, ambiguity and the past performance of motion hypotheses generated by the matches, to estimate the probability that a feature match is correct. At every stage the best correspondences are chosen to generate a hypothesis. This method will therefore only spend time on poor or ambiguous matches when the best correspondences have proven themselves to be unsuitable. The result is a system that is able to operate efficiently on ambiguous data and is suitable for implementation on devices with limited computing resources. [2010 BMVC Abstract] [2010 BMVC Paper] [2010 BMVC Slides]

Deterministic Sample Consensus with Multiple Match Hypotheses
McIlroy, Rosten, Taylor and Drummond

This paper proposes a deterministic scheme for selecting correspondences from feature matching to generate motion hypotheses. The method combines matching scores, ambiguity and the past performance of motion hypotheses generated by the matches, to estimate the probability that a feature match is correct. At every stage the best correspondences are chosen to generate a hypothesis. This method will therefore only spend time on poor or ambiguous matches when the best correspondences have proven themselves to be unsuitable. The result is a system that is able to operate efficiently on ambiguous data and is suitable for implementation on devices with limited computing resources.

[2010 BMVC Abstract] [2010 BMVC Paper] [2010 BMVC Slides]


	Reconstruction from Uncalibrated Affine Silhouettes McIlroy and Drummond This work addresses the problem of model building from multiple affine silhouette views of an object in an uncontrolled environment such as an aircraft in flight. Each pair of silhouette views provides two outer epipolar tangency constraints on the relative motion between the cameras. For a scaled orthographic camera model with six degrees of freedom we show that it is possible to recover structure and motion from six or more silhouette views by solving the outer epipolar tangency constraints simultaneously. [More] [2009 BMVC Abstract] [2009 BMVC Paper] This work won best student poster at BMVC 2009.

Reconstruction from Uncalibrated Affine Silhouettes
McIlroy and Drummond

This work addresses the problem of model building from multiple affine silhouette views of an object in an uncontrolled environment such as an aircraft in flight. Each pair of silhouette views provides two outer epipolar tangency constraints on the relative motion between the cameras. For a scaled orthographic camera model with six degrees of freedom we show that it is possible to recover structure and motion from six or more silhouette views by solving the outer epipolar tangency constraints simultaneously. [More]

[2009 BMVC Abstract] [2009 BMVC Paper] This work won best student poster at BMVC 2009.


	Stability of the Delaunay Triangulation McIlroy and Fitzgibbon Delaunay triangulation is invariant to similarity transformations, but not to affine or perspective transformations and therefore may seem a poor choice for use in wide-baseline matching. This experiment investigates the stability of the Delaunay algorithm for a synthetic scene undergoing wide-baseline changes in viewpoint. We show that Delaunay triangulation is surprisingly stable. [Delaunay Video]

Stability of the Delaunay Triangulation
McIlroy and Fitzgibbon

Delaunay triangulation is invariant to similarity transformations, but not to affine or perspective transformations and therefore may seem a poor choice for use in wide-baseline matching. This experiment investigates the stability of the Delaunay algorithm for a synthetic scene undergoing wide-baseline changes in viewpoint. We show that Delaunay triangulation is surprisingly stable.

[Delaunay Video]

Contact Information

Email:	Paul McIlroy (pmm33)
Address:	Department of Engineering. University of Cambridge Trumpington Street, Cambridge CB2 1PZ
Telephone:	+44 1223 332 748
Fax:	+44 1223 332 662

| Cambridge University | CUED | MIL |