Estimating the Parameters of the K Distribution in the Intensity Domain
Abstract
This paper reviews a number of different moment-based methods for estimating the parameters of the K distribution. The K distribution is a model for the statistics of synthetic aperture radar (SAR) imagery that is formed from the product of two independent distributions, one representing the radar cross-section, and the other representing speckle that is a characteristic of coherent imaging. A method for synthesizing correlated K-distributed random fields is reviewed. Moments of intensity, log intensity and amplitude statistics are used to compute measures which are inverted to give estimates of the mean and order parameters of the K distribution. A method based on the normalized logarithm of intensity is shown to give the best performance, confirming the results of Oliver (1993) and Blacknell (1994). The effect of correlation upon order parameter estimation and its role in the K distribution's underlying component pertaining to the radar cross-section is briefly considered. This work will be used to development target detection algorithms to find targets of interest in SAR imagery.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 01, 1999
- Accession Number
- ADA368069
Entities
People
- Nicholas J. Redding