Research Papers of Ciarán Ó Conaire

• SenseCam Image Localisation\\using Hierarchical SURF Trees
  Ciarán Ó Conaire, Michael Blighe and Noel E. O'Connor
  Multimedia Modelling 2009
  The SenseCam is a wearable camera that automatically takes photos of the wearer's activities, generating thousands of images per day. Automatically organising these images for efficient search and retrieval is a challenging task, but can be simplified by providing semantic information with each photo, such as the wearer's location during capture time. We propose a method for automatically determining the wearer's location using an annotated image database, described using SURF interest point descriptors. We show that SURF out-performs SIFT in matching SenseCam images and that matching can be done efficiently using hierarchical trees of SURF descriptors. Additionally, by re-ranking the top images using bi-directional SURF matches, location matching performance is improved further.
  PDF Version of Paper
  
  
• Unsupervised Feature Selection for Detection Using Mutual Information Thresholding
  Ciarán Ó Conaire and Noel E. O'Connor,
  WIAMIS 2008 - 9th International Workshop on Image Analysis for Multimedia Interactive Services, Klagenfurt, Austria, 7-9 May 2008
  This paper proposes a method for unsupervised selection of features for detecting important events in a surveillance context. While traditional feature selection requires manually annotated ground truth to choose the best features, we examine the possibility of exploiting the redundancy between a pair of independent data sources for selecting good detection features. Building on our prior work on mutual information thresholding, we show that strong agreement between data sources indicates strong detection performance. Experimental tests, combining visual and audio data, show that the best performing features can be automatically selected by taking advantage of the common information shared by the sensors.
  PDF
  
  
• Detector adaptation by maximising agreement between independent data sources
  Ciarán Ó Conaire, Noel E. O'Connor and Alan F. Smeaton,
  IEEE International Workshop on Object Tracking and Classification Beyond the Visible Spectrum 2007
  Traditional methods for creating classifiers have two main disadvantages. Firstly, it is time consuming to acquire, or manually annotate, the training collection. Secondly, the data on which the classifier is trained may be over-generalised or too specific. This paper presents our investigations into overcoming both of these drawbacks simultaneously, by providing example applications where two data sources train each other. This removes both the need for supervised annotation or feedback, and allows rapid adaptation of the classifier to different data. Two applications are presented: one using thermal infrared and visual imagery to robustly learn changing skin models, and another using changes in saturation and luminance to learn shadow appearance parameters.
  Illustrative Videos and Results
  PDF
  
  
• Thermo-Visual Feature Fusion for Object Tracking Using Multiple Spatiogram Trackers
  Ciarán Ó Conaire, Noel E. O'Connor, Alan Smeaton
  [Journal of Machine Vision and Applications, 11 May 2007]
  In this paper, we propose a framework that can efficiently combine features for robust tracking based on fusing the outputs of multiple spatiogram trackers. This is achieved without the exponential increase in storage and processing that other multimodal tracking approaches suffer from. The framework allows the features to be split arbitrarily between the trackers, as well as providing the flexibility to add, remove or dynamically weight features. We derive a mean-shift type algorithm that allows efficient object tracking with very low computational overhead. We especially target the fusion of thermal infrared and visible spectrum features as the most useful features for automated surveillance applications. Results are shown on multimodal video sequences clearly illustrating the benefits of combining multiple features using our framework.
  Video object-tracking experiments
  PDF
  
  
  
• AN IMPROVED SPATIOGRAM SIMILARITY MEASURE FOR ROBUST OBJECT LOCALISATION
  Ciarán Ó Conaire, Noel E. O'Connor, Alan F. Smeaton
  [IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP)]
  Spatiograms were introduced as a generalisation of the commonly used histogram, providing the flexibility of adding spatial context information to the feature distribution information of a histogram. The originally proposed spatiogram comparison measure has significant disadvantages that we detail here. We propose an improved measure based on deriving the Bhattacharyya coefficient for an infinite number of spatial-feature bins. Its advantages over the previous measure and over histogram-based matching are demonstrated in object tracking scenarios.
  view PDF
  Illustrative Videos
  
  
• Organising a Daily Visual Diary Using Multi-Feature Clustering
  Ciarán Ó Conaire, Noel E. O'Connor, Alan F. Smeaton, Gareth J. F. Jones
  [Electronic Imaging 2007]
  The SenseCam is a prototype device from Microsoft that facilitates automatic capture of images of a person's life by integrating a colour camera, storage media and multiple sensors into a small wearable device. However, efficient search methods are required to reduce the user's burden of sifting through the thousands of images that are captured per day. In this paper, we describe experiments using colour spatiogram and block-based cross-correlation image features in conjunction with accelerometer sensor readings to cluster a day's worth of data into meaningful events, allowing the user to quickly browse a day's captured images. Two different linear time algorithms are detailed and evaluated for SenseCam image clustering.
  view PDF
  CDVP's main SenseCam page
  
  
• Multispectral Object Segmentation and Retrieval in Surveillance Video
  Ciarán Ó Conaire, Noel O'Connor, Eddie Cooke, Alan Smeaton
  [ICIP 2006]
  This paper describes a system for object segmentation and feature extraction for surveillance video. Segmentation is performed by a dynamic vision system that fuses information from thermal infrared video with standard CCTV video in order to detect and track objects. Separate background modelling in each modality and dynamic mutual information based thresholding are used to provide initial foreground candidates for tracking. The belief in the validity of these candidates is ascertained using knowledge of foreground pixels and temporal linking of candidates. The Transferable Belief Model is used to combine these sources of information and segment objects. Extracted objects are subsequently tracked using adaptive thermo-visual appearance models. In order to facilitate search and classification of objects in large archives, retrieval features from both modalities are extracted for tracked objects. Overall system performance is demonstrated in a simple retrieval scenario.
  view PDF
  
  
• Comparison of Fusion Methods for Thermo-Visual Surveillance Tracking
  Ciarán Ó Conaire, Noel O'Connor, Eddie Cooke, Alan Smeaton
  [International Conference on Information Fusion (FUSION 2006)]
  In this paper, we evaluate the appearance tracking performance of multiple fusion schemes that combine information from standard CCTV and thermal infrared spectrum video for the tracking of surveillance objects, such as people, faces, bicycles and vehicles. We show results on numerous real world multimodal surveillance sequences, tracking challenging objects whose appearance changes rapidly. Based on these results we can determine the most promising fusion schemes.
  view PDF
  
  
• DETECTION THRESHOLDING USING MUTUAL INFORMATION
  Ciarán Ó Conaire, Noel O'Connor, Eddie Cooke, Alan Smeaton
  presented at VISAPP 2006
  In this paper, we introduce a novel non-parametric thresholding method that we term Mutual-Information Thresholding. In our approach, we choose the two detection thresholds for two input signals such that the mutual information between the thresholded signals is maximised. Two efficient algorithms implementing our idea are presented: one using dynamic programming to fully explore the quantised search space and the other method using the Simplex algorithm to perform gradient ascent to significantly speed up the search, under the assumption of surface convexity. We demonstrate the effectiveness of our approach in foreground detection (using multi-modal data) and as a component in a person detection system.
  view PDF
  
  
• Background Modelling in Infrared and Visible Spectrum Video for People Tracking
  O Conaire C, Cooke E, O'Connor N, Murphy N and Smeaton A.F.
  International Conference on Computer Vision and Pattern Recognition, San Diego, CA, 20-25 June 2005.
  view PDF
  
  
• Fusion of Infrared and Visible Spectrum Video for Indoor Surveillance
  O Conaire C, Cooke E, O'Connor N, Murphy N and Smeaton A.F.
  WIAMIS 2005 - 6th International Workshop on Image Analysis for Multimedia Interactive Services, Montreux, Switzerland, 13-15 April 2005.
  view PDF
  
  

Technical Reports and Documents

• appendix to Thermo-Visual Feature Fusion for Object Tracking Using Multiple Spatiogram Trackers
  This report contains details of two theoretical analyses that arose from previous work in object tracking using multiple feature spatiograms, as well as some discussion on how the work could be extended to cope with changing object properties. The first analysis concerns the models of feature spatial distribution that are made when fusing multiple spatiogram trackers and explain how the use of simple Gaussian distributions for each bin does in fact lead to more flexible distribution models. The second part of this report concerns the similarity measure that is used to compare two spatiogram models. A new similarity measure is derived that more closely resembles a probability measure and its relationship with the Bhattacharyya coefficient is discussed. We also discuss how the object model should be updated to cater for changing conditions.
  
  
  
• PhD Register - Transfer Report
  While traditional image and video processing focus on extracting knowledge from data of a single modality, such as visual spectrum or thermal infrared video, this report investigates the benefits and challenges of capturing and analysing multimodal video. It specifically targets the two modalities of visible spectrum and thermal infrared video. A novel capture device has been developed to capture and align (both temporally and spatially) video sequences captured in both modalities. Two publications in international conferences, to date, demonstrate the effectiveness of multimodal analysis in its robustness to lighting variations and background camou.age of tracked objects. The challenge of multimodal fusion is to obtain optimal benefits from the joint use of multiple modalities, namely thermal infrared and visible spectrum video in this project. Applications include 24- hour automatic surveillance systems, people tracking-and-counting in commercial centres, human-machine interfaces and industrial automation applications.
  view PDF
  
  
• Efficient Euler-Number Thresholding
  Technical Report, Aug 2005
  Euler thresholding is a non-parametric dynamic thresholding method, useful for many computer vision tasks. It requires the calculation of a graph relating thresholds to Euler numbers and performing a Rosin unimodal threshold calculation on the graph. This report details how this can be done in time complexity O(N +T ) where N is the number of pixels in the image and T is the number of thresholds used.
  view PDF