| EPSRC Reference: |
GR/J64542/01 |
| Title: |
OBJECT RECOGNITION TECHNIQUES FOR SYNTHETIC APERTURE RADAR |
| Principal Investigator: |
Delves, Professor L |
| Other Investigators: |
|
| Researcher Co-Investigators: |
|
| Project Partners: |
|
| Department: |
Statistics and Computational Mathematics |
| Organisation: |
University of Liverpool |
| Scheme: |
Standard Research (Pre-FEC) |
| Starts: |
17 May 1994 |
Ends: |
16 May 1997 |
Value (£): |
129,605
|
| EPSRC Research Topic Classifications: |
| RF & Microwave Technology |
|
|
| EPSRC Industrial Sector Classifications: |
|
| Related Grants: |
|
| Panel History: |
|
|
Summary on Grant Application Form |
|
The identification of small features in synthetic aperture radar (SAR) images. SAR images are extremely (100%) noisy and these features are typically two or three pixels in size, thus conventional image analysis techniques are unsuitable. This project will develop algorithms to identify such features, with a library of routines forming the practical side of the project and publications in recognised journals and collections representing the theoretical aspect.Progress:Specific problem selection. It is a intrinsic need of the project to use all available information to solve a specific problem, including particular contextual information, since there is little information contained in one or two pixels. We therefore expect the algorithms developed to be highly problem-specific. In discussion with DRA Malvern, the first problem selected was the identification of a row of approximately equally spaced reflectors; runway lights and electricity pylons form practical applications. Construction of a test bed system to allow ideas to be easily investigated. Since it would be necessary to try many different ideas during the projects lifetime, it was decided from an early stage that a versatile test bed system would be extremely useful. The system that has been constructed adopts a visual programming paradigm, thereby simplifying the chaining of operations that are to be performed on the SAR images.Development of an algorithm to detect runway lights. Such an algorithm is now running, though still subject to further development. Firstly, the SAR image is preprocessed, with a directed, acyclic graph (DAG) being the result. This DAG is then passed to the lines of points detection algorithm, which retains any points which form part of a line, according to user-supplied parameters (for example, the minimum number of points in a line). The preprocessing phase begins with a local thresholding step, which selects points significantly brighter than their locality. Connected points are then aggregated into regions. The size of each region is tested, with regions which fall outside a user-supplied range being discarded. The remaining regions are replaced by points (at their centres of mass), which are then connected (within a user-defined distance) to form a DAG. To locate lines of points, two points are selected and the distance and angle between them calculated. These are then used to predict where the next point will fall. If a point is found in such a position (to a certain tolerance), then the same process is repeated with the second of the two original points and this third point. This is repeated until no point is found in the predicted position. The user is then informed of the presence of this line of points. This process is repeated for every pair of connected points within the DAG (actually optimisation is performed). A refinement of this algorithm has been developed, whereby points can be omitted from a line and the line can still be identified. In such a case, a point which fails to be located is treated as existing if this allows the algorithm to find another point in the predicted position past that.
|
|
Key Findings |
|
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
|
|
Potential use in non-academic contexts |
|
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
|
|
Impacts |
|
| Description |
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk |
| Summary |
|
| Date Materialised |
|
|
|
|
Sectors submitted by the Researcher |
|
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
|
| Project URL: |
|
| Further Information: |
|
| Organisation Website: |
http://www.liv.ac.uk |