Maximum Likelihood Estimation of Compound-Gaussian Clutter and Target Parameters
Abstract
The compound-Gaussian model is often used in radar signal processing to describe the heavy-tailed clutter distribution. The important problems in compound-Gaussian clutter modeling are choosing the texture distribution and estimating its parameters. Many texture distribution models have been proposed 1, 2 and their parameters were typically estimated using the (statistically suboptimal) method of moments, see 2. In this paper, we develop maximum likelihood (ML) methods for jointly estimating target and clutter parameters in compound-Gaussian clutter. In particular, we estimate (i) the complex target amplitudes, (ii) covariance matrix of the speckle component, and (iii) the texture-distribution parameters. Several existing texture models are considered: (i) gamma (leading to the well-known K clutter distribution 1, 2), (ii) lognormal, and (iii) Weibull. Motivated by the robust regression model in 3, we also develop a complex multivariate t distribution model for the clutter. We utilize the expectation-maximization (EM) algorithm to estimate the unknown parameters. Numerical integration is typically needed to compute the conditional expectations in the expectation (E) step of the EM algorithm; here, we employ the Gauss quadratures to perform this integration. Interestingly, the proposed complex multivariate t distribution model does not require numerical integration, allowing for remarkably simple estimation and detection algorithms. We will also compute Cramer-Rao bounds (CRBs) for the unknown parameters and demonstrate the performances of the proposed methods via numerical simulations.
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 20, 2004
- Accession Number
- ADA433621
Entities
People
- Aleksandar Dogandzic
- Arye Nehorai
- Jian Wang
Organizations
- Iowa State University