Sea Clutter Distribution Modeling: A Kernel Density Estimation Approach

TL;DR

This paper introduces a kernel density estimation framework for modeling sea clutter distributions, eliminating the need for prior assumptions and significantly improving accuracy and target detection performance.

Contribution

It develops a non-parametric modeling approach with closed-form bandwidth solutions and a fast iterative algorithm, outperforming traditional parametric models.

Findings

Reduces sea clutter modeling error by about 100 times.

Improves target detection probability by up to 36%.

Provides a flexible, prior-free modeling framework.

Abstract

An accurate sea clutter distribution is crucial for decision region determination when detecting sea-surface floating targets. However, traditional parametric models possibly have a considerable gap to the realistic distribution of sea clutters due to the volatile sea states. In this paper, we develop a kernel density estimation based framework to model the sea clutter distributions without requiring any prior knowledge. In this framework, we jointly consider two embedded fundamental problems, the selection of a proper kernel density function and the determination of its corresponding optimal bandwidth. Regarding these two problems, we adopt the Gaussian, Gamma, and Weibull distributions as the kernel functions, and derive the closed-form optimal bandwidth equations for them. To deal with the highly complicated equations for the three kernels, we further design a fast iterative bandwidth selection algorithm to solve them. Experimental results show that, compared with existing methods, our proposed approach can significantly decrease the error incurred by sea clutter modeling (about two orders of magnitude reduction) and improve the target detection probability (up to $36\%$ in low false alarm rate cases).