This thesis concerns the analysis and processing of sea clutter from a Multiband Pulsed Radar - a land based research system operated by the British Defence Evaluation and Research Agency. This radar serves as a model for a class of Multi Function Radars (MFR) that offer extensive computer controlled adaptive operation. A fast Sequential Edge Detector (SED) is formulated which, accounting for locally exponential speckle, allows the spatial inhomogeneity within a scene to be segmented. This simultaneously identifies high intensity areas and the noise dominated shadowed regions of the scene using an adaptively sized analysis window. The high resolution data is thus shown to contain discrete scatterers which exist in addition to the compound modulation from the wave surface. The discrete component means the measured statistics cannot be considered homogenous or stationary. This is crucial for high resolution MFR as a priori information can no longer be relied upon when viewing a scene for the first time in order to make a detection decision. Considering the returns to be discrete in nature leads to a potential Doppler detection scheme operable at low velocities within the clutter spectrum. A physically motivated test statistic, termed persistence, is demonstrated based upon the lifetime of scattering events determined via the Continuous Wavelet Transform. When operated in coastal regions at low resolution, strong returns from the land-sea interface (edges) are expected which will seriously degrade the performance of radar detection models tuned to homogenous scenes. Explicit operational bounds are determined for the strength of these edges which show that simultaneous operation of an edge detector is required when assessing compound statistics such as the K-distribution using typical texture estimators. Additionally a method for accurately determining the N-sum PDF of K-distributed statistics within noise is constructed using a numerical inverse Laplace transform. The SED is also applied to Synthetic Aperture Sonar data to detect the large shadows cast by targets rather than their point intensity.
| Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:392121 |
| Date | January 2001 |
| Creators | Davidson, Glen |
| Publisher | University College London (University of London) |
| Source Sets | Ethos UK |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
| Source | http://discovery.ucl.ac.uk/1382011/ |
Page generated in 0.0019 seconds