Image Understanding
A number of members are active in the Image understanding group within Signal Processing and Communications Laboratory. Prof.
Nick Kingsbury is the head of the group.
Faculty
Post-doc
Research Students
Recent past members of the group
- Ee Sin Ng (2012)
- James Nelson (2011)
- Tao Hong (2011)
- Pashmina Bendale (2010)
- Ryan Anderson (2007)
- Julian Fauqueur (2006)
Background
The work in this area is largely based on new image analysis tools that have been developed from our earlier work on the Dual-Tree Complex Wavelet Transform (DT-CWT). This transform has been shown to model the early stages of human vision quite closely and it has some desirable properties, not normally present in conventional wavelet transforms, such as strong directional selectivity and shift invariance of the subband basis functions. It also forms an approximately tight frame, that preserves signal energy in the transform domain.
Using the DT-CWT, we have recently developed new feature-point detectors and descriptors, which allow matching of image features in rotationally invariant ways. In addition we are able to use the complex wavelet coefficients efficiently to extract accurate motion fields that are highly consistent with true object motion. Combining these two approaches results in relatively efficient methods for analysing video data of scenes such as traffic at road junctions. In the Sustainable Road Freight project, this will allow us to monitor traffic at large numbers of junctions simultaneously, and develop highly detailed statistical models of journeys and road use. These then will permit much more accurate prediction of road congestion than is currently feasible.
Projects
Some of the current projects in the group:
- Cambridge toy cars dataset
- Object recognition, tracking
- Centre for Sustainable Road Freight ( CSRF) - traffic monitoring and prediction of congestion ( project 2C).
Relevant Papers
Key papers, relevant to activities of this group are:
- N Kingsbury: "Rotation-invariant local feature matching with complex wavelets", Proc. European Conference on Signal Processing (EUSIPCO), Florence, 4-8 Sept 2006, paper #1568982135. (PDF)
- J Fauqueur, N Kingsbury and R Anderson: "Multiscale keypoint detection using the dual-tree complex wavelet transform", Proc. IEEE Conference on Image Processing, Atlanta, GA, 8-11 Oct 2006. (PDF)
- S K Pang, J D B Nelson, S J Godsill and N Kingsbury: "Video Tracking Using Dual-Tree Wavelet Polar Matching and Rao-Blackwellised Particle Filter", EURASIP Journal on Advances in Signal Processing, vol. 2009, Article ID 620404, 13 pages, 2009. (PDF)
- P Bendale, W Triggs and N G Kingsbury: "Multiscale keypoint analysis based on complex wavelets", British Machine Vision Conference 2010, Aberystwyth, Wales, 31 Aug - 3 Sept, 2010. (PDF)
- J D B Nelson and N G Kingsbury: "Enhanced shift and scale tolerance for rotation invariant polar matching with dual-tree wavelets", IEEE Trans on Image Processing, vol 20, no. 3, Mar 2011, pp 814-821. (PDF)
Additional information may be available on the individual web-pages of group members.
