Parameter Estimation of K-Distributed Clutter Based on Fuzzy Inference and Gustafson-Kessel Clustering
The detection performance of maritime radars is restricted by the unwanted sea echo or clutter which is the vector sum of scattering from the sea surface. The echo is noise-like and is expected from a set of randomly moving scatters. Although the number of these target-like data is small, they may c...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English Malay |
| Published: |
2008
|
| Subjects: | |
| Online Access: | http://psasir.upm.edu.my/id/eprint/7868/1/ABS---__FK_2008_59.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | The detection performance of maritime radars is restricted by the unwanted sea echo or clutter which is the vector sum of scattering from the sea surface. The echo is noise-like and is expected from a set of randomly moving scatters. Although the number of these target-like data is small, they may cause false alarm in maritime radar and perturb the target detection. K-distribution is known as the best fitted probability density function for the radar sea clutter. The accurate and fast parameter estimation of K-distribution for small number of sea clutter radar data is crucial task to avoid irreparable disasters.
A novel approach to estimate the parameters of K-distribution based on fuzzy inference has been proposed in the thesis. Takagi-Sugeno Kang (TSK) model has been chosen since human knowledge is unavailable to be captured, whereas the sea clutter for specific parameter can be easily generated. GK- clustering has been used in order to identify the membership function of the antecedent parts. Least Square Method has been utilized to estimate the parameter of the K-distribution, which is represented in the consequent part of the fuzzy inference system. For a real-time implementation of the proposed method, vectorized programming technique has been implemented. In comparison with the conventional methods, this technique has less computational complexity, needs lesser time to train and estimates faster than any existing methods. Since the method is clustering based, some kind of pre-knowledge (rough estimation) is naturally stored in the structure of the TSK-fuzzy system and Least Square provides a mechanism to fine tune the consequent parameters.
The novelty of the proposed method is the incorporation of the clustering (as a pre-estimator) with the estimation process. The resultant estimator then overcomes the bottleneck of the existing methods and is capable of handling even a small number of data. |
|---|