Click here for more photo gallery.

LATEST ACHIEVEMENTS (updated: 11.01.2007)

1. I will leave Cambridge University (for good? I don't know yet) for a visting scholar position in Hong Kong by the end of January 2007 for a number of projects involving Bayesian methods. Should you need to contact me you may try my existing email (kfn20@cam.ac.uk) or my gmail (pseudo1237@gmail.com).
2. We have developed a joint online particle filter for tracking variable number of manoeuvring extended targets using VRPF in a sensor network. A conference paper and a journal paper are being prepared. More details will be given shortly.
3. We have successully integrated the VRPFs, MRF, and a data-dependent
   importance sampling method to obtain an excellent tracking algorithm for multiplt manoeuvring targets. We essentially utilise latest observations to generate better state particles (forward and side thrusts), which is found to have improved the overall target tracking performance when compared with the case when the dynamical prior function is used alone. A conference paper about this proposed method has been accepted to NSSPW 2006. (04.2006)
4. We are now working on the sensor registration problem where the positions
   of the deployed sensors are not precisely known during target detection and estimation. We propose to use a SIR+MCMC method to jointly estimate the target states and the unknown sensor positions, given that a reference sensor with its precise positions is provided. More details will be posted on the web once available. A conference paper about this proposed method has been accepted to NSSPW 2006. (04.2006)
5. We have modified the VRPF methods to enable efficient target tracking for
   multiple manoeuvring targets with multiple sensors in the absence of multiple models. We combine the Independent Partition (IP) tracking with the Markov Random Field (MRF) motion prior to enable efficient tracking of interacting and non-interacting multiple targets in the VRPF framework. A conference paper to EUSIPCO 2006 has been accepted. You may see the updated results here. (01.2006)
6. We have successfully implemented the variable rate particle filters on
   tracking for multiple manoeuvring targets with multiple sensors. This algorithm does not require multiple-models to track manoeuvring targets. A conference paper to CAMSAP 2005 has been submitted and can be downloaded from this web site. You may see the updated results here
7. We also have evaluated our tracking algorithm on a set real-life data
   (Whitworth) provided by our sponsors. The convincing results not only have impressed them, but also invited us to provide them a tailor-made program in Matlab & C++ for their tracking applications. This project is underway and more news will be given. (02.2006)

LATEST PUBLICATIONS

1. Models and algorithms for tracking of manoeuvring objects using variable rate particle filters, IEEE Large Scale Dynamical Systems Workshop (accepted, subject to revision. To appear in April 2007.)
2. Online Multisensor-Multitarget Detection and Tracking Using A Markov Random Field Motion Model, IEEE AES (submitted for review 2006)
3. A hybrid method for online tracking of a variable number of targets, IEEE AES (accepted, subject to revision)
4. Wideband Signal Processing Using MCMC Methods, IEEE SP 2005

1. On tracking applications using variable rate particle filters, NSSPW 2006
2. Efficient Variable Rate Filters For Tracking Manoeuvring Targets Using An MRF-based Motion Model, EUSIPCO 2006
3. Multiple and extended object tracking with Poisson spatial process and variable rate filters, CAMSAP 2005
4. Online Multisensor-Multitarget Detection and Tracking, AES 2006
5. Tracking variable number of targets using Sequential Monte Carlo Methods, SSP 2005
6. A review of recent results in multiple target tracking, ISPA 2005
7. Online Multitarget Detection and Tracking Using Sequential Monte Carlo Methods, Fusion 2005